Online gasoline blending with EPA Complex Model for predicting emissions
Stefan JANAQI1(), Mériam CHÈBRE2, Guillaume PITOLLAT3
1. Ecole des Mines d’Alès, LGI2P, Parc G. Besse, 30035 Nîmes, France 2. TOTAL SA, 24, crs Michelet, 92069 PARIS LA DEFENSE Cedex, France 3. 3Manufacturing Engineering, BECKMAN COULTER, Georgia Institute of Technology, Marseille, France
The empirical Complex Model developed by the US Environmental Protection Agency (EPA) is used by refiners to predict the toxic emissions of reformulated gasoline with respect to gasoline properties. The difficulty in implementing this model in the blending process stems from the implicit definition of Complex Model through a series of disjunctions assembled by the EPA in the form of spreadsheets. A major breakthrough in the refinery-based Complex Model implementation occurred in 2008 and 2010 through the use of generalized disjunctive and mixed-integer nonlinear programming (MINLP). Nevertheless, the execution time of these MINLP models remains prohibitively long to control emissions with our online gasoline blender. The first objective of this study is to present a new model that decreases the execution time of our online controller. The second objective is to consider toxic thresholds as hard constraints to be verified and search for blends that verify them. Our approach introduces a new way to write the Complex Model without any binary or integer variables. Sigmoid functions are used herein to approximate step functions until the measurement precision for each blend property is reached. By knowing this level of precision, we are able to propose an extremely good and differentiable approximation of the Complex Model. Next, a differentiable objective function is introduced to penalize emission values higher than the threshold emissions. Our optimization module has been implemented and tested with real data. The execution time never exceeded 1 s, which allows the online regulation of emissions the same way as other traditional properties of blended gasoline.
If Season= Winter, then y(RVP)=8.7. If x(E300)>95 then y(E300)=95. If x(SUL)<10 then y(SUL)=10, z(SUL)=x(SUL)–10. If x(SUL)>450, then y(SUL)=450, z(SUL)=x(SUL)–450. If x(OLE)<3.77 then y(OLE)=3.77, z(OLE)=x(OLE)–3.77. If x(OLE)>19 then y(OLE)=19, z(OLE)=x(OLE)–19. If x(ARO)<10 then y(ARO)=10, z(ARO)=–8. If x(ARO)<18 then y(ARO)=18, z(ARO)= x(ARO)–18. If x(ARO)>36.8 then y(ARO)=36.8, z(ARO)= x(ARO)–36.8.
1
Balas E (2010). Disjunctive programming. In: 50 Years of Integer Programming 1958–2008: From the Early Years to the State-of-the-Art, 283–340
2
Cason W W (1997). The illusory flexibility of the complex model: a graphical analysis of the implications of the complex model for refinery blending flexibility. In: NPRA Annual Meeting. San Antonio, Texas: NPRA Annual Meeting, 16–18
3
Chèbre M, Creff Y, Petit N (2010). Feedback control and optimization for the production of commercial fuels by blending. Journal of Process Control, 20(4): 441–451 https://doi.org/10.1016/j.jprocont.2010.01.008
4
EPA (2015). Complex emission model.
5
EPA (2016). Template spreadsheet for calculating emissions from combustion plants.
6
Furman K C, Androulakis I P (2008). A novel MINLP-based representation of the original complex model for predicting gasoline emissions. Computers & Chemical Engineering, 32(12): 2857–2876 https://doi.org/10.1016/j.compchemeng.2008.02.002
7
Hirshfield D S, Kolb J A (1994). Minimize the cost of producing reformulated gasoline by integrating EPA’s Complex Model into refinery linear programming (LP) models. NPRA Annual Meeting, 64–67
8
Hirshfield D S, Kolb J A (1997). The economics of gasoline reformulation: refining economics, emissions standards, and the Complex Model. In: NPRA Annual Meeting, San Antonio, USA
9
Jackson J, Vittachi K (1995). CITGO Corpus Christi refinery gasoline blender upgrade. In: NPRA Annual Meeting, San Francisco, USA
10
Janaqi S, Chebre M, Pitollat G (2015). Method and device for monitoring induced properties of a mixture of components, in particular emission properties, 69.
11
Korotney D J (1995). Reformulated gasoline effects on exhaust emissions: Phase III; investigation on the effects of sulfur, olefins, volatility, and aromatics and the interactions between olefins and volatility or sulfur. SAE Special Publications, 179–186
12
Misener R, Floudas C A (2009). Advances for the pooling problem: modeling, global optimization, and computational studies Survey. Applied and Computational Mathematics, 8(1): 3–22
13
Misener R, Gounaris C E, Floudas C A (2010). Mathematical modeling and global optimization of large-scale extended pooling problems with the (EPA) complex emissions constraints. Computers & Chemical Engineering, 34(9): 1432–1456 https://doi.org/10.1016/j.compchemeng.2010.02.014
14
Naman B T (1999). Linear models help refiners develop RFG [(reformulated gasoline)] recipes. Oil & Gas Journal, 97(8): 64–66
15
Nocca J L, Forestiere A, Cosyns J (1994). Diversify process strategies for reformulated gasoline. Fuel Reformulation, 4(5): 18–22
16
Raman R, Grossmann I E (1994). Modelling and computational techniques for logic based integer programming. Computers & Chemical Engineering, 18(7): 563–578 https://doi.org/10.1016/0098-1354(93)E0010-7
17
Treiber S (1998). RFG [(reformulated gasoline)]: the challenge to conventional blending technology. In: Gulf Publishing, eds. Hydrocarbon Processing 2nd International Conference on Process Optimization. Houston: Hydrocarbon Processing 2nd International Conference on Process Optimization, 101–103
18
Trierwiler L D (1995). Representing the simple and complex models in linear programming with respect to reformulated gasoline. In: American Energy Week ’95 “Pipelines, Terminals Storage, and Reformulated Fuels” Conference’ Houston, USA
