**OLEFIN PLANT OPTIMIZATION USING STEADY-STATE MODELS**

PAL, Deepak (SABIC R&T Pvt. Ltd, Plot #81-85 Chikkadunnasandra,Off Sarjapur-Attibele Road, Bangalore 5, 562125, IN)

ILIYAS, Abduljelil (SABIC T&I, P.O. Box 42503, Riyadh, 11551, SA)

**G05B19/042**WO2012021995A1 | 2012-02-23 |

US20070059838A1 | 2007-03-15 | |||

EP2884354A1 | 2015-06-17 | |||

EP1441268A1 | 2004-07-28 | |||

US6816822B1 | 2004-11-09 | |||

US6862562B1 | 2005-03-01 | |||

US6488839B1 | 2002-12-03 | |||

US20100161133A1 | 2010-06-24 | |||

US7930044B2 | 2011-04-19 | |||

US6278899B1 | 2001-08-21 | |||

US7194318B2 | 2007-03-20 | |||

US7213006B2 | 2007-05-01 |

EISELE P ET AL: "Up olefin plant profits by steady state simulation", HYDROCARBON PROCESSING,, 1 August 1980 (1980-08-01), pages 95 - 99, XP001306491

CLAIMS 1. A process for plant optimization, comprising: a) selecting one or more decision variables, screening one or more objective functions, and one or more system parameters; b) determining the one or more system parameters at a time-step; screening the one or more objective functions c) optimizing the one or more decision variables by using steady state models under plant constraints; d) performing dynamic simulation to update the one or more system parameters for a subsequent time-step; and e) repeating steps c) and d) with updated system parameters for a predetermined runtime. 2. The process of claim 1, wherein the decision variable is feedstock variation, and the system parameters are flowrate and composition. 3. The process of claim 1, wherein the decision variable is furnace operation, and the system parameters are conversion, coil/furnace outlet temperature, selectivity, pressure drop, and coke deposition. 4. The process of claim 1, wherein the decision variable is process constraints, and the system parameters are furnace throughput, recovery selection, energy consumption, and impact on sustainability. 5. The process of claim 1, wherein the simplified mass balance model comprises a detailed furnace model and a simplified recovery section, where the simplified recovery section is simulated using component splitters. 6. The process of claim 1, wherein the rigorous model comprises a detailed furnace model and rigorous separation model, wherein the rigorous separation model includes compression, distillation, and refrigeration. 7. A process for plant optimization, comprising: a) selecting one or more decision variables, screening one or more objective functions, and one or more system parameters; b) determining base conditions and values of the one or more system parameters at a time-step; c) using the base conditions as values of the one or more decision variables; d) optimizing steady state yields based on a simplified mass balance model to provide optimized values to the one or more decision variables within certain limits; e) evaluating the one or more objective functions and plant constraints with the optimized values (under the defined limits) using a rigorous model; f) optimizing the one or more objective functions under plant constraints by repeating steps c), d), and e) after updating the base conditions with optimized values; g) performing dynamic simulation to update the one or more system parameters for a subsequent time step; and h) repeating steps c), d), e), f), and g) with updated system parameters for a predetermined runtime. 8. The process of claim 7, wherein the decision variable is feedstock variation, and the system parameters are flowrate and composition. 9. The process of claim 7, wherein the decision variable is furnace operation, and the system parameters are conversion, coil/furnace outlet temperature, selectivity, pressure drop, or coke deposition. 10. The process of claim 7, wherein the decision variable is process constraints, and the system parameters are furnace throughput, recovery selection, energy consumption, and impact on sustainability. 11. The process of claim 7, wherein the simplified mass balance model comprises a detailed furnace model and a simplified recovery section, where the simplified recovery section is simulated using component splitters. 12. The process of claim 7, wherein the rigorous model comprises a detailed furnace model and rigorous separation model, wherein the rigorous separation model includes compression, distillation, and refrigeration. 13. A process for plant optimization, comprising: a) selecting one or more decision variables; b) determining base conditions and coke thickness of a furnace at a time-step; c) taking base conditions as values of the one or more decision variables; d) optimizing steady state yields based on a simplified mass balance model to provide optimized values to the one or more decision variables; e) evaluating plant profits with the optimized values using a rigorous model; f) optimizing the plant profits under plant constraints by repeating steps c), d), and e) after updating the base conditions with the optimized values; g) performing dynamic furnace simulation to update the coke thickness for a subsequent time-step; h) repeating steps c), d), e), f), and g) with the updated coke thickness for a predetermined runtime. 14. The process of claim 13, wherein the simplified mass balance model comprises a detailed furnace model and a simplified recovery section, where the simplified recovery section is simulated using component splitters. 15. The process of claim 13, wherein the rigorous model comprises a detailed furnace model and rigorous separation model, wherein the rigorous separation model includes compression, distillation, and refrigeration. |

CROSS-REFERENCE TO RELATED APPLICATIONS [0001] This application claims the benefit of priority of U.S. Provisional Patent Application No. 62/382,610, filed September 1, 2016, which is hereby incorporated by reference in its entirety.

FIELD

[0002] The presently disclosed subject matter relates to a process for optimization of olefin plant operations, using a multi -tiered approach that simultaneously addresses issues related to profit, energy, and equipment constraints.

BACKGROUND

[0003] Olefins, or alkenes, refer to any unsaturated hydrocarbon chain that contains at least one carbon-carbon double bond. Olefins are used in the manufacture of virtually all consumer products made with chemicals or plastics.

[0004] In order maximize profits during the production of olefin in olefin plants, process optimization models are needed.

[0005] Process models are generally used to predict, control and optimize processes and can be categorized by two model types: steady-state models and dynamic models. In either case, an optimization model is a mathematical process where data related to a system or process is analyzed in order to determine optimal parameter sets for operation of the process. For example, the optimization model disclosed herein provides a method to optimize the profits of an olefin plant by analyzing a number of process parameters.

[0006] Certain methods for optimization of chemical processes, including those for olefin plants exist and have been implemented in the art.

[0007] U.S. Patent Publication No. 2010/0161133 discloses a dynamic optimizer for chemical processes including a maximum feed calculator that performs maximum feed calculations, and a feed coordinator that specifies or implements control strategies. This publication teaches the optimization of production of a chemical plant, such as an olefin plant, using steady state models, but does not disclose predictions of time dependent parameters such as coke thickness.

[0008] U.S. Patent No. 7,930,044 discloses a steady state optimization method incorporating dynamic variance correction in order to improve the controllability of a steady state optimization control system. This system does not teach the use of actual reaction and coking kinetics for product yields and coke formation respectively.

[0009] U.S. Patent No. 6,278,899 discloses an on-line optimizer for optimizing operations of a plant with respect to predetermined operation parameters, including both steady state models and dynamic models. The steady state models are optimized for gain as opposed to an accurate prediction and the dynamic models provide accurate prediction of the system. This system does not teach using dynamic model predictions as input for steady state optimization processes and does not teach using phenomenological/physical model to simulate the steady state and dynamic models.

[0010] U.S. Patent No. 7,194,318 discloses modular formulation of model predictive controllers and teaches the integration of steady state optimization results with dynamic controllers to improve operation stability. It does not teach the use of changing operation parameters in order to increase product yields.

[0011] U.S. Patent No. 7,213,006 discloses neural network based control systems for online optimization of processes using non-linear models for steady state behavior and linear models for dynamic behavior of the processes. This system further discloses that steady state models are optimized for gain as opposed to an accurate prediction, and the dynamics models are what provide accurate predictions of the system. This system does not use actual reaction kinetics for yield prediction and coke formation.

[0012] Typically, those optimization models, which have previously been implemented in olefin plants, use either simplified linear programming models, or use a steady state approach as a way to reduce the costs of complex computations involving multiple optimization variables. In order to create a sustainable optimization model over the entire furnace run length, however, it is important to assess the impact on energy and equipment loadings. To do this, a complete olefin plant simulation including all furnaces, hydrogenation reactors and recovery section must be executed, which creates a highly non-linear problem. Because typical olefin plant optimization models use only linear models or a steady state approach, one common problem is that these models lack the ability to address issues related to revenue, energy, equipment constraints and the dynamic nature of optimization simultaneously.

[0013] There is a need for a model based optimization process for olefin plants that can deal with multiple variables and non-linear optimization in order to address issues related to revenue, energy, equipment constraints simultaneously.

SUMMARY OF THE DISCLOSED SUBJECT MATTER

[0014] The presently disclosed subject matter provides a multi -tiered process for plant optimization. In particular, the present disclosure provides a process for optimization of an olefin plant.

[0015] In certain embodiments, a process for optimization of an olefin plant can include tiered pseudo-steady state optimization methods to simulate dynamic plant operations. The process can further include selecting one or more decision variables, screening one or more objective functions, and one or more system parameters. The process can further include determining the one or more system parameters at a time step and screening the one or more objective functions. The process can further include optimizing the one or more decision variables by using a steady state model under plant constraints. For example, but not by way of limitation, the process can include the use of steady state models to optimize both the yields of furnaces and recovery sections, as well as overall plant profits at a given time-step. [0016] In certain embodiments, the process can further include the use of dynamic furnace models to predict time-dependent system parameters. For example but not by way of limitation, the process can include dynamic models to predict coke thickness. In certain embodiments, the process can further include inputting coke thickness values obtained from dynamic modeling into a steady state model for optimization at another step. In some embodiments, the process may include multiple repeats of the optimization process, updating system parameters for a predetermined run time.

[0017] The presently disclosed subject matter further provides a process for olefin plant optimization where base conditions of certain system parameters may be used as decision variables. In certain embodiments, the process can include selecting one or more decision variables, screening one or more objective functions, and one or more system parameters. In certain embodiments, the process can further include determining base conditions and values of one or more system parameters at a given time-step. The process can further include using the determined base conditions as values of the optimization decision variables. The process can further include optimizing steady state yields based on a simplified mass balance model to provide optimized values to the decision variables within certain limits. The process can further include evaluating the process objective functions and plant constraints with the optimized values using a rigorous model. The process can further include updating the base conditions with optimized values, and inputting these optimized base condition values into a simplified mass balance in order to optimize steady state yields and optimize the one or more objective functions under plant constraints. The process may further include performing a dynamic simulation in order to update the system parameters for a subsequent time step. In some embodiments, the process may include multiple repeats of the optimization process, updating system parameters for a predetermined run time.

BRIEF DESCRIPTION OF THE DRAWINGS

[0018] FIG. 1 is a schematic diagram depicting an exemplary overview of the different modules which make up the disclosed subject matter.

[0019] FIG. 2 is a schematic diagram depicting an exemplary overview of the disclosed subject matter.

[0020] FIGS. 3A and 3B provide schematic diagrams depicting exemplary overviews of the two sets of models that are described in the disclosed subject matter, where 3a depicts a mass balance model and 3b depicts a rigorous simulation model.

[0021] FIGS. 4A and 4B provide schematic diagrams depicting examples of the tiered approach that is described in the disclosed subject matter, where FIG. 4A depicts Tier I, and FIG. 4B depicts Tier II.

[0022] FIG. 5 provides the data for ethane recycle with or without dynamic optimization as described in Example 2. The y-axis shows ethane recycle (tph)x-axis over time (log-scale).

[0023] FIG. 6 provides the data on impact on COT when running the optimizer dynamically as described in Example 2. The y-axis shows COT (°C) over time (log-scale).

DETAILED DESCRIPTION

[0024] The presently disclosed subject matter provides a multi -tiered process for plant optimization. In particular, the present disclosure provides a multi-tiered process for optimization of an olefin plant that simultaneously addresses issues related to profit, energy, and equipment constraints.

[0025] For the purpose of illustration and not limitation, FIG. 1 is a schematic diagram depicting an exemplary overview of the different modules, which make up the disclosed multi-tiered optimization approach.

[0026] As shown in FIG. 1, in certain embodiments, the multi-tiered optimization approach 100 may include various sub-modules, including a furnace simulation model 101, an optimization mode 102, and a recovery section model 103. In certain embodiments, the furnace simulation model 101 has the ability to determine optimal process conditions for maximizing revenue under certain give process constraints. The furnace simulation model 101 may further have the ability to be operated in dynamic mode and may be able to predict the rate of coke formation in the olefin plant at any given time step. In certain embodiments, the recovery section model 103 may have the capability to determine overall energy costs and to evaluate equipment constraints. The optimization model, 102 is further outlined in FIG. 2.

[0027] For the purpose of illustration and not limitation, FIG. 2 is a schematic diagram depicting an exemplary overview of the disclosed multi-tiered optimization process 200.

[0028] As shown in FIG. 2, the first step in the optimization process 200 includes development of a base case 201, which may be developed and validated using design and plant data. In certain embodiments, the base case 201 may be developed using two sets of models. For the purpose of illustration and not limitation, FIGS. 3A and 3B provide schematic diagrams depicting examples of these two sets of models.

[0029] As seen in FIG. 3A, the first model developed under base case 201 is a simplified mass balance model 301. In certain embodiments, the simplified mass balance model 301 can include a detailed furnace model 302 and a simplified recovery section 303 simulated using component splitters. In certain embodiments, the detailed furnace model 302 can be based on known cracker kinetics. The simplified mass balance model 301 is not able to solve heat and pressure balances. In certain embodiments however, the simplified mass balance model 301 may be configured to run yield optimization runs and to evaluate multiple optimization scenarios. In certain embodiments, the simplified mass balance 301 may be executed using Aspen Plus SM Model software, 308. Further features of this mass balance model are shown in FIG. 3B.

[0030] As shown in FIG. 3B, the second model developed under base case 201 is a rigorous simulation model 304. In certain embodiments, rigorous simulation model 304 can include a detailed furnace model 305, and a rigorous separation section 306 including compression, distillation and refrigeration. The rigorous simulation model 304 can be used to estimate overall energy requirements and impact on equipment loadings 307. In certain embodiments, the rigorous simulation model 304 may be simulated in a sequential way using Aspen Plus SM Model software, 308. In certain embodiments, rigorous simulation model 304 can be linked with an activated energy analysis, exchangers design and rating, and other tools which are compatible with the sequential modular approach of Aspen Plus 308.

[0031] As shown in FIG. 2, the second step of the disclosed multi-tiered optimization process 200 includes various optimization scenarios 202 which may be evaluated using the simplified mass balance model 301 shown in FIG. 3A. For each optimization scenario 202, an optimization function, decision variables, and suitable constraints are defined. For example, but not by way of limitation, Table 1 provides an example optimization scenario.

Table 1 : Optimization Scenarios

[0032] As shown in Table 1, in an example scenario, the optimization functions can include maximizing ethylene and maximizing total profits. The decision variables can include coil/furnace outlet temperature and steam to oil/hydrocarbon ratio. In other embodiments, for example, but not by way of limitation, decisions variables used in optimization scenario 202 can includes conversion, number of furnaces, or feed composition to each furnace. In certain embodiments, due to the nature of the simplified mass balance model 301, it may be possible to quickly evaluate various optimization scenarios 202. For example, but not by way of limitation, by using a standard desktop configuration, such as an Intel 17 4790, 8 GB RAM, it may be possible to execute one optimization scenario 202 in the range of 30 seconds to two minutes. Therefore, multiple optimization scenarios 202 can be evaluated in a very short amount of time. The executed optimization scenario 202 can then be used to determine the optimization conditions for an entire plant, as shown in FIG. 2.

[0033] As shown in FIG. 2, the third step of the disclosed multi-tiered optimization process 200 includes optimization of the whole plant 203. In certain embodiment, the rigorous simulation model shown in FIG. 3B may be used to carry out whole plant simulations 203. Simulations of the whole plant however can be very complicated, and in order to use the rigorous approach 304 to simulate the entire plant, a multi -tiered approach must be used. In certain embodiments, this multi-tiered approach can include two tiers, Tier I and Tier II.

[0034] In certain embodiments, Tier I of the multi-tiered approach for simulating the entire plant is a steady state optimization. During Tier I, operations are considered to be at a steady state. The base case as defined in step one of FIG. 2 may be used as the starting point of optimization in Tier I. In certain embodiments, Tier I can make use of optimization capabilities of the simplified model and can carry out energy computations using the rigorous model. The impact of coke formation on conversion, coil/furnace outlet temperature, and selectivity however, is not evaluated as part of Tier I. Thus, for furnaces that are controlled using coil/furnace temperature, running the furnace at temperatures which have been optimized for a certain steam/oil hydrocarbon ratio may result in a violation of equipment constraints, and will put an additional load on the recovery section. In order to account for the impact of coke formation over time, a second tier, Tier II may be implemented. In certain embodiments, Tier II of the multi-tiered approach for simulating the entire plant may make use of the Tier I steady state optimization at each time step. In certain embodiments, in Tier II, each furnace can be simulated individually. Further, at each time step in Tier II, an average coking reaction rate across the furnace can be obtained using an in-house coking model or other similar tools. For example and not by way of limitation, it may be possible to use other tools, such as Neural Networks, in order to obtain the rate of the coking reaction.

[0035] For the purpose of illustration and not limitation, FIGS. 4A and 4B provide schematic diagrams depicting examples of this tiered approach, where FIG. 4A depicts Tier I, and FIG. 4B depicts Tier II.

[0036] As shown in FIG. 4A, in certain embodiments, Tier I may include a series steps. The first step, 401, can include a run case with a rigorous model with base conditions. In this step, a profit is calculated using the equation: PROFIT(O) = Revenue(O) - Fuel Costs (0). In certain embodiments, at step 401, the base case equipment loading will also be computed. The second step, 402 can include relaxing the constraints on decision variables by a very small margin. For example, but not by way of limitation, in the case of coil/furnace temperature, the variable may be relaxed by 2-5 °C, where the base case was 5 °C. The third step, 403 can include running a steady state yield optimizer and computing new optimized parameters. The fourth step, 404, can include running the entire case in a rigorous model with the optimized process conditions, and calculating the profit using the equation: Pr(i+1) = PROFIT (i+1) = Revenue (i+1) - Fuel Cost (i+1). In certain embodiments, at step 404, the equipment loading and impact on constraints may also be evaluated. After running the rigorous model of step 404, if the PROFIT(i+l) is greater than PROFIT(O) and the equipment constraints are not violated, then the process should be looped back to step 402 and the base case parameters should be changed to optimization results. As shown in FIG. 4A, the steps of Tier I should be looped continuously until the results of the rigorous model at step 404 show that the PROFIT(i+l) is no longer greater than PROFIT(O). Once this result is reached, Tier I is complete.

[0037] As shown in FIG. 4B, in certain embodiments, Tier II may include a series of steps. The first step, 405, can include determining whether the temperature (t) is greater than a maximum temperature, (t max). If t is not greater than t max, then the process moves to the next step, 406, where whole-plant optimization is performed using Tier I. At this step, new optimized parameters are computed, and the averaged coking reaction rate in each furnace is computed. Once the whole plant optimization is complete using Tier 1, the process then moves to the next step, 407 where the new coke thickness is computed, the EOR criteria of each furnace is checked, and the temperature is incrementally increased. The process then loops back to step 405 until the temperature is greater than t max. Once t is greater than t max, Tier II is complete.

[0038] As shown in FIG. 2, after the whole plant has been optimized, the fourth step of the disclosed multi-tiered optimization process 200 includes optimization in real time 204. EXAMPLES

[0039] The following example is merely illustrative of the presently disclosed subject matter and should not be considered as a limitation in any way.

Example 1:

[0040] This example illustrates an implementation using the multi-tier approach as it is disclosed herein.

[0041] In this example, a mix of ethane and propane furnaces was used. The cracking severity of each furnace was as follows:

Conversion in Ethane furnace = 70%

Conversion in Propane furnace = 90%

[0042] In this example, the base conditions used as a starting point are showing in Table 2 and Table 3.

Table 2: Design Data

Table 3 : Furnace Input Data

[0043] In this example, various optimization scenarios were evaluated using the steady state yield optimization model. The scenarios are outlined in Table 4, and the results of the steady state yield optimizations are shown in Table 5.

Table 4: Optimization Scenarios

Table 5 : SS Yield optimization Results

[0044] In this example, the global constraints used in the steady state yield optimization were:

• Max HC Flow Rate to Furnaces:

o Ethane Furnace = 32 tph

o Propane Furnace = 47 tph

• Min/Max Coil/Furnace Outlet Temperature:

o All furnaces = 830 °C - 870 °C

• Recycle Flowrates:

o C2 Recycle = 50 tph

o C3 Recycle = 23 tph

• Min. Production

o Propylene = 18 tph

[0045] In this example, Case 3 was chosen as the optimization scenario for which the whole plant optimization was to be carried out.

[0046] The base case was chosen as the starting point for the Tier I optimization. The coil/furnace outlet temperature for C2 base case was 846 °C. The cracked gas compressor should be less than 42.7 MW. In the first pass, the steady state yield optimization was carried out by incrementally varying the coil/furnace outlet temperature of the steady state yield optimization. Based on the bounds, new optimized parameters were obtained as shown in Table 6.

Table 6: Simulation Cases -Varying Bounds

[0047] In this example, it was observed that coil/furnace outlet temperatures were relaxed incrementally by 5 °C, and at step 4 the system was at an optimum overall global coil/furnace outlet temperature.

[0048] The next step in this example was to run the whole plant optimization at each step and check the energy requirement and equipment constraints. The products and energy requirements at each step of Tier I optimization are shown in Table 7.

Table 7: Products and Energy Requirement

The optimization benefits at each tier step are shown in Table 8. Table 8: Optimization Benefits at Each Tier Step

[0050] As shown in Table 8, there was an impact of energy on the overall profits achieved by performing the optimization. Additionally, at each step in the optimization process, the equipment loadings and plant constraints could be evaluated, which was not possible using steady state yield optimization alone.

[0051] In this example, the CGC power constraint was 42.5 MW which was violated in step 3. Therefore, the most realistic optimization results are those which correspond to step 2, where an overall additional profit of $9.2 million/year is shown, as compared to the $14.94 million/year predicted using the steady state yield optimization in Table 5.

Example 2:

[0052] This example illustrates the impact of dynamic optimization using the Tier II approach as it is disclosed herein.

[0053] FIG. 5 shows the impact of steady state yield optimization in an experiment where the optimization is carried out only at the start of the run, vs. at each time step.

[0054] As the coke gets deposited on the cracking tube surface, in order to achieve the same conversion, a higher coil/furnace outlet temperature is required. In a case where the coil/furnace outlet temperature is kept constant (steady state optimization), conversion is reduced with time due to coke formation. This causes an increase in recycle flowrate, and thus increases the load on the recovery section, which is less than optimal.

[0055] When the steady state yield is run dynamically using the multi-tier approach, the coil/furnace outlet temperature is increased with the furnace run time in order to ensure that conversion is kept constant and the recycle flow is kept under bounds. This is illustrated in FIG. 6

* * *

[0056] In addition to the various embodiments depicted and claimed, the disclosed subject matter is also directed to other embodiments having other combinations of the features disclosed and claimed herein. As such, the particular features presented herein can be combined with each other in other manners within the scope of the disclosed subject matter such that the disclosed subject matter includes any suitable combination of the features disclosed herein. The foregoing description of specific embodiments of the disclosed subject matter has been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the disclosed subject matter to those embodiments disclosed.

[0057] It will be apparent to those skilled in the art that various modifications and variations can be made in the systems and methods of the disclosed subject matter without departing from the spirit or scope of the disclosed subject matter. Thus, it is intended that the disclosed subject matter include modifications and variations that are within the scope of the appended claims and their equivalents.

[0058] Various patents and patent applications are cited herein, the contents of which are hereby incorporated by reference herein in their entireties.

**Previous Patent:**CRYOGENIC AIR SEPARATION METHOD FOR PRODUCING OXYGEN AT HIGH PRESSURES

**Next Patent: SYSTEMS AND METHODS FOR POWER PRODUCTION INCLUDING ION TRANSPORT COMPONENTS**