Edwin Zondervan

Optimal design of a multi-product biorefinery system

Abstract In this paper we propose a biorefinery optimization model that can be used to find the optimal processing route for the production of ethanol, butanol, succinic acid and blends of these chemicals with fossil fuel based gasoline. The approach unites transshipment models with a superstructure, resulting in a Mixed Integer Non-Linear Program (MINLP). We consider a specific problem based on a network of 72 processing steps (including different pretreatment steps, hydrolysis, fermentation, different separations and fuel blending steps) that can be used to process two different types of feedstock. Numerical results are presented for four different optimization objectives (maximize yield, minimize costs, minimize waste and minimum fixed cost), while evaluating different cases (single product and multi-product).

Energy optimization in the process industries : unit commitment at systems level

In previous work, the unit commitment problem has been formulated as a non-convex MINLP, which is computationally expensive to solve. To circumvent this problem, we reformulate in this paper the problem as a convex MIQCP and derive the relevant constraints using propositional logic. We consider a specific problem based on a network of gas and oil fired power generators. Numerical results are presented using several methods (e.g. CPLEX for MIQCP, and DICOPT, SBB for MINLP). It is shown that proposed convex MIQCP reformulation can be solved much faster than previous non-convex models reported in the literature.

Design of a sustainable biorefinery

In this paper we propose a biorefinery optimization model that can be used to find the optimal processing route for the production of ethanol, butanol, succinic acid and blends of these chemicals with fossil fuel based gasoline. The approach unites transshipment models with a superstructure, resulting in a Mixed Integer Non-Linear Program (MINLP). We consider a specific problem based on a network of 72 processing steps (including different pretreatment steps, hydrolysis, fermentation, different separations and fuel blending steps) that can be used to process two different types of feedstock. Numerical results are presented for four different optimization objectives (maximize yield, minimize costs, minimize waste and minimum fixed cost), while evaluating different cases (single product and multi-product).

Optimizing the tactical planning in the Fast Moving Consumer Goods industry considering shelf-life restrictions

Abstract This paper addresses the optimization of the tactical planning for the Fast Moving Consumer Goods industry using an MILP model. To prevent unnecessary waste and missed sales, shelf-life restrictions are introduced using three methods. The direct method tracks the age of products directly. While it provides optimal solutions, it is computationally inefficient. The indirect method forces products to leave inventory before the end of their shelf-life. It obtains solutions within a few percent of optimality. Moreover, compared to the direct method, the computational time was on average reduced by a factor 32. The hybrid method models the shelf-life directly in the first and indirectly in the second storage stage. It obtains near optimal solutions and, on average, reduces the required computational time by a factor 5 compared to the direct method. Cases containing up to 25, 100, and 1000 SKUs were optimized using the direct, hybrid and indirect method respectively.

Integrated Operation and Design of a Simulated Moving Bed Reactor

Abstract In this work we use a simulated moving bed reactor (SMBR) model that can be used to optimize operational and design parameters of the process. An SMB unit contains several columns with specific lengths that are connected sequentially. Feed and desorbent are supplied and extract and raffinate can be withdrawn from the system. The process of supply/withdrawal is cyclic, which makes optimization of operation and design a challenging task. We will discuss our model in this contribution and show the preliminary results that we obtain for an isomerization process of glucose into fructose. In addition we discuss how the model can be incorporated in a conceptual framework that can be used to integrate the operation and design of the SMBR.

Simultaneous Optimization of Planning and Scheduling in an Oil Refinery

Abstract In earlier work we have developed and tested a scheduling model [1] in the AIMMS software. In this follow-up contribution we will develop a planning model. Next we will identify the information flow between scheduling model and the planning model. Lastly we will integrate the two models in a straight-forward fashion using run-modes and rolling-horizon methodology. With a case study we prove that the overall strategy aids systematic integration of the two levels, which allows fast optimization of the short-term as well as the long-term decisions without significantly affecting the quality of the solution

Modelling of packed bed adsorption columns for the separation of green tea catechins

ABSTRACT Green tea is a rich source of catechins, which when purified have a high economic value as they can be used as a supplement in several products, to increase their health benefits. Catechins are regarded as desired components with several applications in a variety of areas such as foods, cosmetics and pharmaceuticals. A multicomponent sorption model has been developed for the separation of catechins from liquid tea streams, with macroporous resins in a packed bed column. Two commercially available food grade resins were considered: Amberlite XADHP and Diaion HP20. For the desorption step, two food grade solvents are used: water and ethanol. The adsorption and desorption behaviour is subsequently mathematically described with one-dimensional axial dispersed plug flow model that can accurately simulate the dynamics of the solvent swing sorption columns. The model parameters were regressed from experimental data. Five components are modelled in the competitive sorption: the main four catechins present in green tea and caffeine. The model was used for the process design and optimization for the recovery of catechins from green tea.

Integrated Product and Process Design for the Optimization of Mayonnaise Creaminess

To optimize the sensory product attributes by changing the processing conditions or product recipe was the main objective of this research. A mathematical framework was built containing process and sensory models. First, a typical mayonnaise production line was modeled. The line consisted of two mixing steps; in mixer 1 the objective was to prepare an oil-in-water emulsion; while in mixer 2 the required product specifications had to be reached. The droplet size and emulsion viscosity were coupled in the processing model for mixer 2. The physicochemical emulsion properties were subsequently correlated to the sensory attributes with a Neural Network. This allowed us to estimate panel scores on sensory attributes. Second, an optimization case study was formulated with the objective to increase mayonnaise creaminess while fixing the oil concentration to a minimal value of 0.65 w/w. The overall result was that the creaminess could be increased by 22 %, but at the expense of other sensory attributes.

Black tea cream effect on polyphenols optimization using statistical analysis

Abstract Black tea cream formation is an inhibitor for the polyphenols separation since it decreases the amount of available polyphenols. Four factors that are considered to have an impact in the amount of tea cream and polyphenols availability are studied: temperature, amount of solids, pH and amount of EDTA. By using a design of experiments instead of a one-factor-at-a-time, additional information such as interaction effects can be obtained. The objective is to determine the optimum combination range for the factors that minimize the cream formation, while maximizing the amount of polyphenols in the clear phase. Statistical analysis is used to determine which factors significantly influence the responses and to generate polynomial models. This is a very effective tool and it indicates that EDTA is the only non-relevant factor. The optimization results in a 37% increase in the yield of theaflavins and a 20% increase in the yield of catechins.

Integrating Planning and Scheduling in an Oil Refinery with a Rolling Horizon Approach

Decisions in an oil refinery are made at three levels: planning, scheduling and control. Existing facilities have to be operated close to their maximum capacity, while continuously responding to cost fluctuations. In many of the currently reported planning models each decision level has its own model and is used separately. Integration of these models will lead to solutions closer to the optimum than modeling the decision levels separately. However, simply combining the levels into one model would lead to very large models that yet cannot be solved within reasonable time. To overcome the computational time issue a rolling horizon approach is proposed in this work.