Research Article An Energy Efficiency Evaluation Method ...Research Article An Energy Efficiency...
Transcript of Research Article An Energy Efficiency Evaluation Method ...Research Article An Energy Efficiency...
Research ArticleAn Energy Efficiency Evaluation Method Based on EnergyBaseline for Chemical Industry
Dong-mei Yao,1 Xin Zhang,2 Ke-feng Wang,3 Tao Zou,2 Dong Wang,1 and Xin-hua Qian4
1School of Environmental Science and Technology, Dalian University of Technology, Dalian 116024, China2Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang 110016, China3School of Chemical Engineering, Dalian University of Technology, Dalian 116024, China4Fushun Petrochemical Branch Company, PetroChina, Fushun 113004, China
Correspondence should be addressed to Tao Zou; [email protected] and Dong Wang; [email protected]
Received 17 February 2016; Accepted 9 August 2016
Academic Editor: Giuseppe Vairo
Copyright ยฉ 2016 Dong-mei Yao et al. This is an open access article distributed under the Creative Commons Attribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
According to the requirements and structure of ISO 50001 energy management system, this study proposes an energy efficiencyevaluation method based on energy baseline for chemical industry. Using this method, the energy plan implementation effect inthe processes of chemical production can be evaluated quantitatively, and evidences for system fault diagnosis can be provided.This method establishes the energy baseline models which can meet the demand of the different kinds of production processes andgives the general solving method of each kind of model according to the production data. Then the energy plan implementationeffect can be evaluated and also whether the system is running normally can be determined through the baseline model. Finally,this method is used on cracked gas compressor unit of ethylene plant in some petrochemical enterprise; it can be proven that thismethod is correct and practical.
1. Introduction
According to the โStatistical Review of World Energyโ whichis revised annually by BP company, the worldโs primaryenergy consumption structure for last ten years is shown inTable 1. Currently, the total annual energy consumption ofthe world is equivalent to about 13 billion tons of oil, andaverage growth rate of energy consumption is 2%per year; thefossil energy such as oil, natural gas, and coal occupies morethan 86% of total consumption which shows that the worldenergy structure is based on fossil fuels. At the same time, theproblems of energy depletion and environmental pollutionhave become increasingly serious. It will take a long time todiscover and develop new energies and renewable energies,so the fossil fuels will be the energy base of human survivaland development in a long period and the problems of fossilfuels can only be alleviated by increasing the energy efficiencynow.
Therefore, energy efficiency in chemical industry orga-nizations has been controlled and systematized by using apowerful instrument which is called energy management
system (EnMS). It can assist the organizations to develop andimplement an energy policy and establish quality objectives,targets, and action plans which take into account legalrequirements and information related to significant energyuse, improve energy efficiency, reduce costs, and improveenvironmental protection and benefits at last. And the EnMSconforming to international standard is based on ISO 50001standard.The ISO 50001 standard is drawn up and publishedby the ISO international organization for standardizationISO/PC 242, energy management committee. The ISO 50001standard can guide the organization to establish EnMS. Someresearches have given the methods about how to establishthe EnMS framework based on ISO 50001 in kinds of indus-tries specifically [1โ5]. For example, an EnMS frameworkproposed by Vikhorev et al. could implement the standardsfor energy data exchange, on-line energy data analysis,performance measurement, and display of energy usage [1].Giacone and Manco regard the whole energy system of a siteas a single matrix equation model, use the model to quantifythe specific energy consumption, and provide support for theenergy benchmarking, budgeting, and targeting; finally the
Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2016, Article ID 7087393, 10 pageshttp://dx.doi.org/10.1155/2016/7087393
2 Mathematical Problems in Engineering
Table 1: The worldโs primary energy consumption structure during last decade.
Year Total primary energy/milliontons of oil equivalent
Share of each energy in primary energy structure/%Oil Natural gas Coal Nuclear energy Hydropower Renewable energy
2005 10557.1 36.1 23.5 27.8 6.0 6.3 /2006 10878.5 35.8 23.7 28.4 5.8 6.3 /2007 11099.3 35.6 23.8 28.6 5.6 6.4 /2008 11294.9 34.8 24.1 29.2 5.5 6.4 /2009 11164.3 34.8 23.8 29.4 5.4 6.6 /2010 12002.4 33.6 23.8 29.6 5.2 6.5 1.32011 12225.0 33.4 23.8 29.7 4.9 6.5 1.72012 12476.6 33.1 23.9 29.9 4.5 6.7 1.92013 12730.4 32.9 23.7 30.1 4.4 6.7 2.22014 12928.4 32.6 23.7 30.0 4.4 6.8 2.5
method is used in the glass and cast iron melting processessuccessfully [2]. Chiu et al. study the case of an electronicparts and components company which has established anintegration-energy-practice model for introducing an ISO50001 energy management system; the results of this studyprove that the model could efficiently meet demands forenergy performance indicators [4].
However, ISO 50001 is only a framework; the researchesmentioned above are also the extension of this frameworkin detail, it is not methodology-based, it cannot implementquantitative evaluation of energy efficiency by ISO 50001only, and this shortage restricts the effective application ofEnMS in the complex chemical industry. So it needs toimplement the quantitative evaluation of energy efficiency,the EnMS will be effective, and then the EnMS can beused to produce a comprehensive understanding of theoverall energy situation, main problems, and the energyconsumption and energy-saving potential and eventuallyimprove energy efficiency and economic benefit. Throughcombining the ISO 50001 standard framework and chemicalbusiness characteristics [6], this research puts forward aquantitative evaluation method of energy efficiency basedon energy baseline for chemical industry, where the energybaseline is a curve that can react to the energy performanceof system or equipment; it is quantitative reference providinga basis for comparison of energy performance. Therefore,the key of this method is to establish the energy baselinemathematical model that can accurately reflect the actualenergy performance in a specified period of time.Then it canimplement the energy efficiency evaluation by comparing thedifferences between the actual energy performance and theenergy performance calculated by energy baseline model.
2. Methods
2.1. ISO 50001 Standard Framework. The purpose of ISO50001 standard is to provide a standard framework whichassists organizations to establish the energymanagement sys-tem and necessary energy management process to improveenergy performance. According to ISO 50001 requirements,the energymanagement system structure is shown in Figure 1.
The ISO 50001 standard is based on Plan-Do-Check-Act (PDCA) continual improvement principle, the details ofthe respective steps are shown in Figure 1, and each step isdivided into technical level and management level as shownin Figure 2.
This study focuses on the technical level of ISO 50001standard. Based on the technical level framework, thisresearch will center on energy baseline and discuss how toestablish energy baseline model that is suitable for differentkinds of chemical production processes. And then energyefficiency evaluationwill be implemented by using the energybaselinemodel and the technical level of the EnMS is ensuredeventually to run as PDCA continual improvement principle.
2.2. Establishment of Energy Baseline Model
2.2.1. Energy Performance Indicators and Impact Factors. Ac-cording to the results of the energy review, it is ensuredthat the energy performance can be measured and impactsthe energy efficiency, energy use, and energy consumption.Then, according to the measured energy input parameters,the energy performance will further be expressed by energyperformance indicators (EnPIs) that can illustrate whetherthe energy performance is good or not. Then, the impactfactors affecting EnPIs are analyzed and selected.
2.2.2. The Form of Energy Baseline Model. The EnPIs andimpact factors can be expressed as the following functionequation:
๐ = ๐ (๐, ๐) + ๐, (1)
where ๐ = (๐1, ๐2, . . . , ๐๐)๐ is an ๐-dimensional columnvector; it represents ๐ EnPIs; ๐ = (๐1, ๐2, . . . , ๐๐)๐ is an๐-dimensional column vector; it represents๐ impact factorsof the EnPIs; ๐(โ ) = [๐1(โ ), ๐2(โ ), . . . , ๐๐(โ )]๐ is a mappingrelation from ๐ to ๐; ๐ is a coefficient vector in the mappingrelations; ๐ = (๐1, ๐2, . . . , ๐๐); ๐๐ (๐ = 1, 2, . . . ,๐) is a randomvariable that obeys the normal distribution whose mean iszero and variance is ๐2.
Mathematical Problems in Engineering 3
Management commitment and ensuring energy policy 4.2/4.3
Energy review: analyze energy use and consumption, identify the areas of significant energy use, and identify & prioritize & record opportunities for improving energy performance 4.4.3
Determine energy performance 4.4.3 Establish access for legal requirements 4.4.2
Energy performance indicator 4.4.5 Legal requirements 4.4.2 Other requirements 4.4.2
Establish energy baseline 4.4.4Other target benchmarks &technical benchmarks 4.4.2
Energy objectives & targets & action plans 4.4.6
Monitoring, measurement, and analysis 4.6.1
Nonconformities & correction action & preventive action 4.6.2 Evaluation of compliance with the requirements 4.6.2
Internal audit of the EnMs 4.6.3
Management review 4.7
Implementation and operation 4.5
Design 4.5.6 Documentation 4.5.4 Training 4.5.2
Energy procure 4.5.7 Operational 4.5.5 Communication 4.5.3
Plan
Do
Check
Act
Continual improvement
Figure 1: Energy management system architecture.
Technical
Plan(i) Energy review (4.4.3)
(ii) Energy baseline (4.4.4)(iii) Energy performance
indicators (4.4.5)
Do(i) Design (4.5.6)
(ii) Energy purchasing (4.5.7)
Check(i) Monitoring (4.6.1)
(ii) Measurement (4.6.1)(iii) Verifying action plans
results (4.4.6)
Act(i) Energy performance and
EnPIs review (4.7.1)
Management
Plan(i) Policy/goals/targets (4.3,
4.4.6)(ii) Resources (4.2.1)
Do(i) Training (4.5.6)
(ii) Communication (4.5.7)(iii) Documentation (4.5.4)(iv) Operational (4.5.5)
Check(i) Internal audit (4.6.3)
(ii) Corrective/preventiveaction (4.6.4)
Act(i) Management review (4.7)
Plan Do
Check Act
Figure 2: Technical level and management level of ISO 50001 standard.
4 Mathematical Problems in Engineering
Depending on the different energy objects, the energybaseline models are divided into three types: (A) empiricalequation or mechanistic model is given directly in the formof (1); (B) mechanistic model is difficult to be obtained andthe energy object is linear; then the energy baseline model isdefined as a linear model; (C) mechanistic model is difficultto be obtained and energy object shows a strong nonlinearcharacter, the energy baseline model is defined as a sum ofnonlinear functions of each impact factor, and each nonlinearfunction of each impact factor is approximated by B-splinedifference function. In the following, the energy baselinemodel forms are given as mentioned by (B) and (C).
(1) Linear Model. Each EnPI is expressed as linear form ofimpact factors, as shown in the following:
๐1 = ๐01 + ๐11๐1 + ๐21๐2 + โ โ โ + ๐๐1๐๐ + ๐1,๐2 = ๐02 + ๐12๐1 + ๐22๐2 + โ โ โ + ๐๐2๐๐ + ๐2,...๐๐ = ๐0๐ + ๐1๐๐1 + ๐2๐๐2 + โ โ โ + ๐๐๐๐๐ + ๐๐.(2)
(2) B-Spline ApproximationModel. For many chemical indus-try processes, it is difficult to give an accurate empirical ormechanistic functional relationship, especially when thereis a strong nonlinear relationship between the EnPIs andthe impact factors. Therefore, general function forms areproposed to approximate various nonlinear relationships.
The EnPIs are the sum of nonlinear relationships of eachimpact factor, as shown in the following:
๐1 = ๐11 (๐1) + ๐21 (๐2) + โ โ โ + ๐๐1 (๐๐) + ๐1,๐2 = ๐12 (๐1) + ๐22 (๐2) + โ โ โ + ๐๐2 (๐๐) + ๐2,...๐๐ = ๐1๐ (๐1) + ๐2๐ (๐2) + โ โ โ + ๐๐๐ (๐๐) + ๐๐.(3)
In (3), each๐๐๐(๐๐) is approximated by a๐-order B-splinefunction. And ๐ is adjusted to adapt to different nonlineardegree models: the higher the nonlinear degree of the modelsis, the bigger ๐ is. It generally can meet most of the needswhen ๐ takes 3 [7โ9].
Firstly, each ๐๐ is divided into ๐ฟ๐ segments evenly, thelength of each segment is โ๐, the segmented point valuesare denoted from small to large as ๐๐0, ๐๐2, . . . , ๐๐๐ฟ๐ , eachsegmented point corresponds to a function value denoted as๐ฝ๐0, ๐ฝ๐2, . . . , ๐ฝ๐๐ฟ๐ , and these function values are the same asthe parameters that need to be solved in theB-spline function.So โ๐, ๐ฟ๐, and the segmented point values are consistent withthe relationship as shown in the following:
โ๐ = max (๐๐) โ min (๐๐)๐ฟ๐ ,๐๐(๐+1) = โ๐ + ๐๐๐.
(4)
The approximate function of B-spline for ๐๐๐(๐๐) in the๐th segment range is as shown in the following equation:
๐๐๐,๐ (๐๐) = ๐โ๐ก=0
๐ฝ๐(๐+๐กโ3) โ ๐บ๐ก,๐(๐๐ โ ๐๐๐โ๐ ) , (5)
where ๐ = 1, 2, . . . , ๐, ๐ = 1, 2, . . . ,๐, ๐ = 3, 4, . . . , (๐ฟ๐ โ 1),๐๐๐ โค ๐๐ โค ๐๐(๐+1), and ๐บ๐ก,๐((๐๐ โ ๐๐๐)/โ๐) is basis function of๐-order B-spline function; make ๐ฅ = (๐๐ โ๐๐๐)/โ๐, ๐ฅ โ [0, 1];then (6) can be deducted:
๐บ๐ก,๐ (๐ฅ) = 1๐!๐โ๐กโ๐ =0
(โ1)๐ ๐ถ๐ ๐+1 (๐ฅ + ๐ โ ๐ก โ ๐ )๐ . (6)
The recursive equation is shown as follows:
๐บ๐ก,0 (๐ฅ) = {{{1 0 โค ๐ฅ โค 10 otherwise,
๐บ๐ก,๐ (๐ฅ) = 1๐ (๐ก + 1 โ ๐ฅ)๐บ๐ก,๐โ1 (๐ฅ)+ 1๐ (๐ โ ๐ก + ๐ฅ)๐บ๐กโ1,๐โ1 (๐ฅ) .
(7)
2.2.3. Model Parameter Solution. The form of the energybaseline model is obtained in Section 4; it needs to solvethe model parameters according to the energy data. ๐พ groupof energy data is given: (๐1๐, ๐2๐, . . . , ๐๐๐, ๐1๐, ๐2๐, . . . , ๐๐๐),๐ = 1, 2, . . . , ๐พ. It can form ๐ โ ๐พ algebraic equations aboutthe model parameters of ๐ by putting the energy data into theenergy baseline model as shown in the following:
๐11 = ๐1 (๐11, ๐21, . . . , ๐๐1, ๐) + ๐11,๐21 = ๐2 (๐11, ๐21, . . . , ๐๐1, ๐) + ๐21,...๐๐1 = ๐๐ (๐11, ๐21, . . . , ๐๐1, ๐) + ๐๐1,๐12 = ๐1 (๐12, ๐22, . . . , ๐๐2, ๐) + ๐12,...๐๐๐พ = ๐๐ (๐1๐พ, ๐2๐พ, . . . , ๐๐๐พ, ๐) + ๐๐๐พ.
(8)
Themodel parameter value is obtained byminimizing theobjective function in the following equation:
๐ฝ (๐)= 12๐พโ๐=1
๐โ๐=1
[๐๐๐ โ ๐๐ (๐1๐, ๐2๐, . . . , ๐๐๐, ๐) โ ๐๐๐]2 . (9)
Equation (9) can be simplified as ๐ โ ๐พ sum of errorsquares of ๐ as shown in the following equation:
๐ฝ (๐) = 12๐พโ ๐โ๐=1
๐2๐ (๐) . (10)
Mathematical Problems in Engineering 5
Generally, (10) is a differentiable function of ๐; then thenecessary condition for ๐ฝ(๐) to obtain the minimum value isshown in the following equation:
๐๐ฝ (๐๐)๐๐๐ = ๐พโ ๐โ๐=1
[๐๐ (๐) ๐๐ (๐)๐๐๐ ] = 0, (11)
where ๐ = (๐1, ๐2, . . . , ๐๐ )๐, ๐ = 1, 2, . . . , ๐ , and these twocases will be discussed according to the different forms of๐๐(๐) of ๐.(1) ๐ as a Linear Relationship. In this case, ๐๐(๐) is a linearexpression of ๐ as shown in the following equation:
๐๐ (๐) = ๐ด๐ โ ๐ + ๐๐0 + ๐๐, (12)
where ๐๐(๐)/๐๐๐ = ๐๐๐; ๐ด๐ = (๐๐1, ๐๐2, . . . , ๐๐๐ ).Then (11) is a linear equation group as shown in the
following equation:
(๐ด๐๐ด)๐ ร๐
โ ๐ = ๐ด๐ (๐ + ๐) , (13)
where
๐ด = [[[[[[[
๐ด1๐ด2...๐ด๐พ๐
]]]]]]](๐พโ ๐)ร๐ ,
๐ = [[[[[[[
๐10๐20...๐(๐พโ ๐)0
]]]]]]](๐พโ ๐)ร1,
๐ = [[[[[[[
๐1๐2...๐(๐พโ ๐)
]]]]]]](๐พโ ๐)ร1.
(14)
Generally,๐ด is nonsingular; then ๐ has a unique solution.Because ๐ is an unknown random perturbation and there is(15), that is,
๐ธ (๐) = ๐ธ [(๐ด๐๐ด)โ1 ๐ด๐ (๐ โ ๐)]= ๐ธ [(๐ด๐๐ด)โ1 ๐ด๐๐] โ (๐ด๐๐ด)โ1 ๐ด๐๐ธ (๐)= (๐ด๐๐ด)โ1 ๐ด๐๐.
(15)
So ๏ฟฝ๏ฟฝ which is the average value of ๐ is the estimate valueof ๐ as shown in the following equation:
๏ฟฝ๏ฟฝ = (๐ด๐๐ด)โ1 ๐ด๐๐. (16)
The linear energy baseline model in (2) and the B-splineapproximation model in (3) and (5) are both linear models ofthe parameters, so the models are solved by this method.
(2) ๐ as a Nonlinear Relationship. When ๐๐(๐1๐, ๐2๐, . . . ,๐๐๐, ๐) is a nonlinear expression of ๐, (11) is nonlinearequation group, the solution would not be unique, therewould be many pole points in (10), and the solution of(11) would not ensure making (10) minimized. Therefore,(10) is minimized directly. And the method combining PSO(Particle SwarmOptimization) andGauss-Newton algorithmis proposed to solve this problem.
PSO algorithm is an evolutionary computation methodbased on themovement of swarms looking for themost fertilefeeding location. It uses a population-based, self-adaptivesearch optimization technique. Therefore, the global bestsolution can be found by simply adjusting the trajectory ofeach individual toward its own best location and towardthe best particle of the entire swarm at each time step. Sothe PSO method is widely used due to its characteristicsof global convergence and simplicity [10โ12]. However, theconvergence speed of PSO is slowed down soon when it isclosed to the global optimal point. So the search is popularabout the solution of this problem. A hybrid optimizationalgorithm (HOA) is put forward to deal with delaminationidentification in laminated composite structures. The HOAis based on combined PSO and SM (simplex method) [10].Ratnaweera et al. discussed two methods to improve theperformance of the PSO based on time-varying accelerationcoefficients (TVAC): first by adding a small perturbation to arandomly selectedmodulus of the velocity vector of a randomparticle by predefined probability and second by discussing akind of self-organizing hierarchical particle swarm optimizerwith TVAC [11].
Gauss-Newton algorithm is a well-known classical iter-ative algorithm to solve nonlinear least squares problems[13]. It uses Taylor series expansion to approximate thenonlinear regression model. Through iteration to correctthe regression coefficient of the nonlinear regression modelcontinuously, the regression coefficient will approximate theoptimum regression coefficient of the nonlinear regressionmodel continuously. Finally the sum of squared residuals tothe original model is minimized. Meanwhile Gauss-Newtonalgorithm is local convergence, and the accuracy and speedof the algorithm depend on selection of the initial value to alarge extent.
So, according to the characteristics of the PSO andGauss-Newton algorithm, the combination of the two methods isproposed to solve above nonlinear optimization problem [14].Firstly, PSO is used to provide a rough solution. Then therough solution is used as the iterative initial value of Gauss-Newton algorithm; the exact minimum value of (10) willbe obtained by finite iterations. The steps using the PSOalgorithm to acquire the rough solution are as follows.
(1) Parameter Setting.The dimension of PSO is๐ , which is thenumber of the model parameters. The number of particles is๐ท; the upper and lower bounds of the particleโs positions are(๐min๐ , ๐max๐ ); the upper and lower limits of particle velocities
6 Mathematical Problems in Engineering
are (Vmin๐ , Vmax๐ ), ๐ = 1, 2, . . . , ๐ . The position of ๐ particle is
expressed as ๐๐ = (๐๐1 , ๐๐2 , . . . , ๐๐๐ ), the particle velocity is ex-pressed as V๐ = (V๐1 , V๐2 , . . . , V๐๐ ), the history optimal positionof the particle ๐ is expressed as ๐๐, and the optimal positionof all the group is expressed as ๐๐, ๐ = 1, 2, . . . , ๐ท. Themaximum weight is ๐คmax, the minimum weight is ๐คmin, andthe maximum number of iterations is ๐max.
(2) Initialization. Make the number of iterations ๐ = 0; theparameters are initialized according to the following:
๐๐๐ (0) = ๐min๐ + (๐max
๐ โ ๐min๐ ) โ rand (1) ,
V๐๐ (0) = Vmin๐ + (Vmax
๐ โ Vmin๐ ) โ rand (1) ,
๐๐ = ๐๐ = ๐1 (0) ,(17)
where rand(1) is a function for generating a random valuethat meets standard normal distribution: ๐ = 1, 2, . . . , ๐ ; ๐ =1, 2, . . . , ๐ท.
(3) Particle Evaluation. Using (9) as the fitness function, thefitness of each particle can be calculated: ๐ฝ๐(๐) = ๐ฝ[๐๐(๐)]. If๐ฝ๐(๐) < ๐ฝ(๐๐), the history optimal position of particle ๐๐ isupdated as ๐๐ = ๐๐(๐); if ๐ฝ๐(๐) < ๐ฝ(๐๐), then the optimalposition of all the particles is updated as ๐๐ = ๐๐(๐).(4) Particles Update.The position and speed of the particle areupdated by the following:
V๐๐ (๐ + 1) = ๐ค โ VV๐๐ (๐) + ๐1 โ rand (1) โ [๐๐ โ ๐๐๐ (๐)]+ ๐2 โ rand (1) โ [๐๐ โ ๐๐๐ (๐)] ,
๐๐๐ (๐ + 1) = ๐๐๐ (๐) + V๐๐ (๐ + 1) ,๐ค = ๐คmax โ ๐๐max
(๐คmax โ ๐คmin) .(18)
(5) Stop Condition. When the fitness function reaches thesetting accuracy or ๐ reaches the maximum number of steps๐max, the iteration stops, and the result of ๐ is ๐๐.
The rough solution๐๐ is within the smaller neighborhoodof the exact optimal solution. Using ๐๐ as the iterative initialvalue of Gauss-Newton algorithm, the solution steps are asfollows.
(1)Make the iteration number ๐ = 0, select the initial valuepoint ๐(0) = ๐๐, and calculate the objective function value asshown in the following equation:
๐ฝ [๐ (0)] = 12๐พโ ๐โ๐=1
๐2๐ [๐ (0)] . (19)
(2) Calculate ๐๐๐(๐) = ๐๐๐[๐(๐)]/๐๐๐, ๐ = 1, 2, . . . , (๐พ โ ๐),and ๐ = 1, 2, . . . , ๐ ; then ๐๐๐(๐) can form the matrix ๐บ(๐) asshown in the following:๐บ (๐)
= [[[[[[[
๐11 (๐) ๐12 (๐) โ โ โ ๐1๐ (๐)๐21 (๐) ๐22 (๐) โ โ โ ๐2๐ (๐)... ... d...๐(๐พโ ๐)1 (๐) ๐(๐พโ ๐)2 (๐) โ โ โ ๐(๐พโ ๐)๐ (๐)
]]]]]]](๐พโ ๐)ร๐ . (20)
The gradient vector of ๐ฝ(๐) is ๐(๐) = [๐1(๐), ๐2(๐), . . . ,๐๐ (๐)]๐, where๐๐ (๐) = ๐พโ ๐โ
๐=1
๐๐ [๐ (๐)] ๐๐๐ [๐ (๐)]๐๐๐ . (21)
(3) The linear equations ๐บ๐(๐) โ ๐บ(๐)๐ = โ๐(๐) of vector ๐are solved, where if the rank of ๐บ๐(๐) โ ๐บ(๐) is ๐ , the equationhas the unique solution ๐(๐) = โ[๐บ๐(๐) โ ๐บ(๐)]โ1 โ ๐(๐); if therank of ๐บ๐(๐) โ ๐บ(๐) is less than ๐ , ๐(๐) = โ๐(๐).
(4) Minimize the linear function group ๐น[๐(๐) + ๐ โ ๐(๐)]of the single variable ๐. The direct linear search method isused to obtain ๐(๐), and ๐(๐) meet ๐น[๐(๐) + ๐(๐) โ ๐(๐)] =min๐๐น[๐(๐) + ๐ โ ๐(๐)]; then the new iteration point and thetarget function value are as shown in the following:
๐ (๐ + 1) = ๐ (๐) + ๐ (๐) โ ๐ (๐) ,๐ฝ [๐ (๐ + 1)] = 12
๐พโ ๐โ๐=1
๐2๐ [๐ (๐ + 1)] . (22)
(5) If the termination condition is satisfied, the iterationwill stop, and ๏ฟฝ๏ฟฝ = ๐(๐+1) is the final solution; otherwisemake๐ = ๐ + 1, and return ๐ step to continue the iteration.
2.2.4. Energy Baseline Model Checking. Through the aboveprocesses, the final energy baseline model is followed:
๐ = ๐ (๐, ๏ฟฝ๏ฟฝ) . (23)
The adjusted decision coefficient is used as the evaluationcriteria to check the obtained model.
Firstly, the estimated values of EnPI are calculated on eachmeasured point:
๏ฟฝ๏ฟฝ๐๐ = ๐ (๐1๐, ๐2๐, . . . , ๐๐๐, ๏ฟฝ๏ฟฝ) ,๐ = 1, 2, . . . , ๐พ, ๐ = 1, 2, . . . ,๐. (24)
Make๐ฆ = (๐11, ๐21, . . . , ๐๐1, . . . , ๐1๐พ, ๐2๐พ, . . . , ๐๐๐พ) ,๏ฟฝ๏ฟฝ = (๏ฟฝ๏ฟฝ11, ๏ฟฝ๏ฟฝ21, . . . , ๏ฟฝ๏ฟฝ๐1, . . . , ๏ฟฝ๏ฟฝ1๐พ, ๏ฟฝ๏ฟฝ2๐พ, . . . , ๏ฟฝ๏ฟฝ๐๐พ) ; (25)
the following statistics can be calculated.Total Square Sum is given as follows:
TSS = (๐ฆ โ ๐ฆ)๐ (๐ฆ โ ๐ฆ) . (26)
Mathematical Problems in Engineering 7
Explain Square Sum is given as follows:
ESS = (๏ฟฝ๏ฟฝ โ ๐ฆ)๐ (๏ฟฝ๏ฟฝ โ ๐ฆ) . (27)
Then the adjusted decision coefficient is as shown in thefollowing equation:
๐ 2 = 1 โ ๐พ ร ๐ โ 1๐พ ร ๐ โ ๐ โ 1 [1 โ (ESSTSS
)2] , (28)
where ๐ 2 is adjusted decision coefficient. The value rangeof ๐ 2 is [0, 1]; the more close to 1, the better to express therelationship between the EnPI and the impact factors for theenergy baseline model. The threshold can be set in advanceaccording to the need; if the adjusted decision coefficient isgreater than the threshold, it is considered that the modelfitting is good; otherwise it is considered that themodel fittingis poor; the model structure or impact factors need to bereselected.
2.3. Energy Efficiency Evaluation Based on Energy Baseline.According to the above sessions, the energy baseline canimplement two functions in EnMS:
(1) Energy baseline can be used as energy performanceevaluation benchmark after the implementation ofenergy plan.
(2) Energy baseline can be used as energy performancemonitoring and system fault diagnosis on the baseof its real-time estimation of energy performanceindicators.
2.3.1. Statistics Related to EnPIs. Firstly, three benchmarkstatistics are provided to judge whether energy performanceachieves the desired standard:
(1) Expected reduced amplitude of EnPIs: ๐๐ท = (๐๐ท1 , ๐๐ท2 ,. . . , ๐๐ท๐)(2) Themean sample deviation of energy baseline model:๐๐ต = (๐๐ต1 , ๐๐ต2 , . . . , ๐๐ต๐)
๐๐ต๐ = 1๐พ๐พโ๐=1
[๐๐๐ โ ๏ฟฝ๏ฟฝ๐๐] . (29)
(3) The standard deviation of sample deviation of energybaseline model:
๐๐ต = (๐๐ต1 , ๐๐ต2 , . . . , ๐๐ต๐) ,๐๐ต๐ = โ 1๐พ โ 1
๐พโ๐=1
[๐๐๐ โ ๏ฟฝ๏ฟฝ๐๐ โ ๐๐ต๐]2,๐ = 1, 2, . . . ,๐.
(30)
Then a period of time to be evaluated is defined, andthen the measured values of EnPIs and impact factors are(๐ฆ1๐, ๐ฆ2๐, . . . , ๐ฆ๐๐, ๐ฅ1๐, ๐ฅ2๐, . . . , ๐ฅ๐๐), ๐ = 1, 2, . . . , ๐. The
measured values of EnPIs can be expressed as the followingform:
๐ฆ๐ = (๐ฆ๐1, ๐ฆ๐2, . . . , ๐ฆ๐๐)๐ , ๐ = 1, 2, . . . ,๐. (31)
Then the impact factors values are substituted into (21) ofenergy baselinemodel in order to obtain the estimated valuesof EnPIs:
๏ฟฝ๏ฟฝ๐ = (๏ฟฝ๏ฟฝ๐1, ๏ฟฝ๏ฟฝ๐2, . . . , ๏ฟฝ๏ฟฝ๐๐)๐ . (32)
Define ๐๐ = (๐๐1, ๐๐2, . . . , ๐๐๐)๐ = ๐ฆ๐ โ ๏ฟฝ๏ฟฝ๐; then themean value and standard deviation of ๐๐ are
๐๐ = 1๐๐โ๐=1
๐๐๐,๐๐ = โ 1๐ โ 1
๐โ๐=1
๐2๐๐.(33)
2.3.2. Energy Performance Evaluation. In order to evaluatethe energy performance during a period of time after theimplementation of the energy plan, the energy baselinemodel is established firstly according to the data before theimplementation of the energy plan; then evaluated valuesof EnPIs through the model in this period are obtained.Through the comparison between the estimated values andthe measured values, the implementation effect of the energyplan can be estimated quantitatively. Specifically, for each๐๐ (๐ = 1, 2, . . . ,๐), ๐๐ is calculated through (33) andcompared with ๐๐ท๐. The evaluation criteria are as follows:
(1) If (โ๐๐) โฅ ๐๐ท๐, the energy management systemachieves or exceeds expected effect; it meets the re-quirements of indicators.
(2) If (โ๐๐) < ๐๐ท๐, the actual reduced amplitude of EnPIis less than expected value, and it does not meetthe requirements; the enterprise needs to review itsenergy again and make a new plan.
2.3.3. Energy Performance Monitoring and System Fault Diag-nosis. Energy efficiencymonitoring and system fault diagno-sis method based on energy baseline are used in the checkingperiod of energy management system. The energy baselinemodel reflects the current state of the system operation;whether the system is normal operation can be judged by it.Specifically, for each ๐๐ (๐ = 1, 2, . . . ,๐), ๐๐ is calculatedthrough (33), and the judging criteria are as follows:
(1) If (๐๐ต๐ โ 3๐๐ท๐) โค ๐๐ โค (๐๐ต๐ + 3๐๐ท๐), the energy man-agement system runs normally.
(2) If ๐๐ < (๐๐ต๐ โ 3๐๐ท๐), the energy baseline model is nolonger suitable for estimating the current state ofEnPI; the model needs to be rebuilt according to thecurrent data.
(3) If ๐๐ > (๐๐ต๐ + 3๐๐ท๐), the energy consumption is toohigh or the system is failure; enterprise needs to makea new plan or system needs maintenance.
8 Mathematical Problems in Engineering
2.4. PDCA Principle Implementation of Energy Baseline. TheEnMS is based on Plan-Do-Check-Act (PDCA) continualimprovement principle, and the energy baseline will beadjusted in different phases to implement different functions:
(1) In Plan phase, energy baseline is established firstlybased on energy history information and is as thebenchmark of the next phase.
(2) In Do phase, the energy baseline is used to evaluatethe difference of the energy performance before andafter the implementation of the energy plan.
(3) In Check phase, the whole system runs smoothly, theenergy baseline needs to be readjusted based on theenergy data after the Do phase, and, at this time,the energy baseline can be used to monitor energyefficiency or diagnose system fault.
(4) In Action phase, the energy efficiency needs to befurther improved to take actions; then it goes intothe Plan phase; the present energy baseline is as thebenchmark for next phase, and the EnMS forms aclosed loop.
3. Results and Discussion
In this section, the cracked gas compression units of the ethy-lene plant in some petrochemical companies are regardedas the energy management object in order to establish theenergy efficiency monitoring and evaluation system accord-ing to the ISO 50001 standard and decrease the energyconsumption. And the specific indicators are to reduceenergy consumption of per unit output more than 5%.
3.1. Cracked Gas Compression Unit Process Principle. Themain equipment of cracked gas compressor unit includes fivestage compressors whose numbers are K-13001, K-13002, K-13003, K-13004, and K-13005; alkali washing tower whosenumber is C-1340; and high pressure depropanizer towerwhose number is C-1365.
In order to prevent compressor surge since the entryof the compressor flow is too small, 3 antisurge lines aredesigned in the cracked gas compressor:
(1) โThree return oneโ: return a portion of the gasdischarged from the third stage compressor to the
suction inlet of the first stage compressor to ensurethat the first three compressorsโ flow is greater thanrequired minimum flow.
(2) โFour return fourโ: return a portion of the gasdischarged from the fourth stage compressor to thesuction inlet of the fourth stage compressor to ensurethat the fourth compressorโs flow is greater thanrequired minimum flow.
(3) โFive return fiveโ: return a portion of the gas dis-charged from the fifth stage compressor to the highpressure propane tower; then the gas of the top ofthe high pressure propane tower returns into the fifthstage compressor.
3.2. Energy Baseline of Cracked Gas Compression Unit. Toimplement advanced process control (APC) system, it isplanned to reduce the return flow of the 4 surge lines toimprove the output of unit energy consumption and achievethe purpose of energy-saving [15โ19].
3.2.1. EnPI and Impact Factors. Energy input for cracked gascompressor unit is ultrahigh pressure steam; product outputfor cracked gas compressor unit is compressed cracked gas,so EnPI is defined as
EnPI = turbine inlet flowK-13005 outlet flow
. (34)
And the impact factors are ๐1 to ๐15 shown in Table 2.
3.2.2. Energy Baseline Model. Since the cracked gas compres-sion unit is working near the set point, it presents linearrelationship between EnPI and impact factors, so the energybaseline model is defined as a linear model as follows:
๐ = ๐0 + 15โ๐=1
๐๐๐๐, (35)
where ๐0 and ๐๐ are model parameters, ๐ is an EnPI, ๐๐ is animpact factor, and ๐ = 1, 2, . . . , 15. Variables defined in detailand parts of the measured data are shown in Table 2.
The model parameters are solved according to the linearmodel solution method; the results are as follows:
๏ฟฝ๏ฟฝ= (0.5244 0.0003 0.0310 โ0.0222 0.0045 โ0.2359 0.0016 โ0.0218 โ0.0191 0.1481 โ0.0001 0.0289 โ0.7438 0.1478 โ0.1323 0.0545)๐ ,๐ 2 = 1 โ 120 โ 1120 โ 16 โ 1 [1 โ (0.2720.283)2] = 0.904.
(36)
๐ 2 is 0.904, which is close to 1 (generally, for the timeseries, it is considered to have a higher fitting accuracy that๐ 2achieves 0.9.), so the model has higher fitting accuracy.
3.3. Energy Performance Evaluation. Through the applicationof the advanced process control on the cracking gas compres-sor unit, the return flow ratio of each surge line is reduced.
Mathematical Problems in Engineering 9
Table 2: Energy performance related variables and parts of measured data.
Name Number Unit Data (per minute)10:00 10:01 10:02 10:03 10:04 10:05 โ โ โ
The compressor speed ๐1 RPM 4078 4067 4025 4070 4059 4012 โ โ โ K-13001 inlet pressure ๐2 KPA 33.3 32.8 32.9 33.2 33.0 33.2 โ โ โ K-13001 outlet temperature ๐3 โC 35.0 34.9 35.4 35.4 35.3 35.2 โ โ โ K-13004 outlet pressure ๐4 โC 75.2 74.8 73.9 75.2 74.9 74.7 โ โ โ K-13004 outlet pressure ๐5 MPA 1.555 1.549 1.540 1.547 1.548 1.549 โ โ โ K-13004 outlet flow ๐6 T/H 241.9 241.9 242.7 245.2 243.2 241.4 โ โ โ K-13005 outlet temperature ๐7 โC 70.9 71.8 70.8 71.3 70.8 71.2 โ โ โ K-13005 outlet pressure ๐8 MPA 3.593 3.634 3.660 3.654 3.639 3.638 โ โ โ Turbine inlet pressure ๐9 MPA 11.48 11.59 11.58 11.54 11.65 11.56 โ โ โ Turbine inlet temperature ๐10 โC 518.2 515.9 515.5 514.8 513.5 515.0 โ โ โ C-1365 tower top temperature ๐11 โC โ21.5 โ21.9 โ21.6 โ21.5 โ21.7 โ21.8 โ โ โ C-1365 tower top pressure ๐12 MPA 1.461 1.440 1.451 1.458 1.463 1.442 โ โ โ K-1300 three return one ๐13 % 7.1 7.2 7.1 7.2 7.1 7.2 โ โ โ K-1300 four return four ๐14 % 6.5 6.5 6.4 6.4 6.5 6.5 โ โ โ K-1300 five return five ๐15 % 6.6 6.6 6.6 6.6 6.6 6.5 โ โ โ K-13005 outlet flow ๐16 T/H 223.7 223.8 223.1 222.8 226.5 226.5 โ โ โ Turbine outlet flow ๐17 T/H 323.1 322.8 320.0 324.0 323.5 322.3 โ โ โ EnPI (๐17/๐16) ๐ โ 1.444 1.442 1.434 1.454 1.428 1.423 โ โ โ
Measured valueEstimated value
1.100
1.150
1.200
1.250
1.300
1.350
1.400
1.450
1.500
EnPI
4:21:00 4:41:00 5:01:00 5:21:00 5:41:00 6:01:004:01:00Time (hh:mm:ss)
Figure 3: EnPI contrast curve before and after implementation ofAPC.
The compression unit is working stably on the point close tothe surge line, and energy consumption of per unit of outputis reduced. The effect analysis of energy-saving is as follows.
The measured data is obtained in two hours after imple-mentation of APC. And the comparison result between theactual measured value and estimated values of EnPI afterimplementing APC is shown in Figure 3.
The actual measured values of EnPI are expressed as๐ฆ = (๐ฆ1, ๐ฆ2, . . . , ๐ฆ120)๐, the estimated values of EnPI cal-culated by the energy baseline model are expressed as ๏ฟฝ๏ฟฝ =
(๏ฟฝ๏ฟฝ1, ๏ฟฝ๏ฟฝ2, . . . , ๏ฟฝ๏ฟฝ120)๐, ๐ = ๐ฆ โ ๏ฟฝ๏ฟฝ is defined, and statistics arecalculated as follows:
๏ฟฝ๏ฟฝ = 1120120โ๐=1
๐ฆ๐ = 1.3686,๐ = 1120
120โ๐=1
๐๐ = โ0.1213.(37)
So the average of reduced amplitude of EnPI is
EnPI reduced amplitude ratio = (โ ๏ฟฝ๏ฟฝ๐ฆ) โ 100= 8.86%.
(38)
Therefore, according to the evaluation criteria of energyefficiency, the establishment of the energy managementsystem and the used technical method complete the expectedindicator to save energy by 5% for the cracked gas compres-sion unit.
4. Conclusions
This study analyzes the energy structure of the current worldand the problems faced and expounds the necessity of energymanagement in chemical enterprises. Based on ISO 50001standard, the structure of energy management system isestablished, and the energy performance evaluation methodbased on energy baseline is proposed. The method firstly
10 Mathematical Problems in Engineering
establishes the baseline models for different kinds of systemsand gives the general solutionmethod for each kind ofmodelsand then illustrates how to use the energy baseline model forenergy performance evaluation and system fault diagnosis.Finally, the application of this method in the cracked gascompression unit of certain petrochemical enterprise makesthe object that energy performance is decreased by 5% cometrue, and the production costs are greatly reduced at the sametime. It can be concluded that this method is correct andpractical.
Competing Interests
The authors declare that they have no competing interests.
Acknowledgments
This research was supported by the National High Technol-ogy Research and Development Program (โ863โ Program)of China (2014AA041802) and the National Natural ScienceFoundation of China (61374112). The authors would like tothank all of the partners for significant contribution in allstages of the research.
References
[1] K. Vikhorev, R. Greenough, and N. Brown, โAn advancedenergy management framework to promote energy awareness,โJournal of Cleaner Production, vol. 43, no. 5, pp. 103โ112, 2013.
[2] E. Giacone and S. Manco, โEnergy efficiency measurement inindustrial processes,โ Energy, vol. 38, no. 1, pp. 331โ345, 2012.
[3] S. A. Ates and N. M. Durakbasa, โEvaluation of corporate en-ergy management practices of energy intensive industries inTurkey,โ Energy, vol. 45, no. 1, pp. 81โ91, 2012.
[4] T.-Y. Chiu, S.-L. Lo, and Y.-Y. Tsai, โEstablishing an integration-energy-practice model for improving energy performance indi-cators in ISO 50001 energy management systems,โ Energies, vol.5, no. 12, pp. 5324โ5339, 2012.
[5] D. B. Goldstein and C. A. Eley, โA classification of building en-ergy performance indices,โ Energy Efficiency, vol. 7, no. 2, pp.353โ375, 2014.
[6] M. Rudberg, M. Waldemarsson, and H. Lidestam, โStrategicperspectives on energymanagement: a case study in the processindustry,โ Applied Energy, vol. 104, pp. 487โ496, 2013.
[7] M. Unser, A. Aldroubi, and M. Eden, โB-spline signal process-ing. I. Theory,โ IEEE Transactions on Signal Processing, vol. 41,no. 2, pp. 821โ833, 1993.
[8] J. R. Nunhez, C. B. B. Costa, and R. Guirardello, โA new splineapproach for data fitting,โ Fluid Phase Equilibria, vol. 220, no. 2,pp. 171โ180, 2004.
[9] H. Park and J.-H. Lee, โB-spline curve fitting based on adaptivecurve refinement using dominant points,โ Computer AidedDesign, vol. 39, no. 6, pp. 439โ451, 2007.
[10] X. Qian, M. Cao, Z. Su, and J. Chen, โA hybrid particleswarmoptimization (PSO)-simplex algorithm for damage iden-tification of delaminated beams,โ Mathematical Problems inEngineering, vol. 2012, Article ID 607418, 11 pages, 2012.
[11] A. Ratnaweera, S. K. Halgamuge, and H. C. Watson, โSelf-organizing hierarchical particle swarm optimizer with time-varying acceleration coefficients,โ IEEE Transactions on Evolu-tionary Computation, vol. 8, no. 3, pp. 240โ255, 2004.
[12] I. Hassanzadeh and S. Mobayen, โController design for rotaryinverted pendulum system using evolutionary algorithms,โMathematical Problems in Engineering, vol. 2011, Article ID572424, 17 pages, 2011.
[13] H. O. Hartley, โThe modified Gauss-Newton method for thefitting of non-linear regression functions by least squares,โTech-nometrics, vol. 3, no. 2, pp. 269โ280, 1961.
[14] S. Zendehboudi, M. A. Ahmadi, A. Bahadori, A. Shafiei, and T.Babadagli, โA developed smart technique to predict minimummiscible pressure-EOR implications,โ The Canadian Journal ofChemical Engineering, vol. 91, no. 7, pp. 1325โ1337, 2013.
[15] U. E. Lauks, R. J. Vasbinder, P. J. Valkenburg, and C. vanLeeuwen, โOn-line optimization of an ethylene plant,โ Comput-ers and Chemical Engineering, vol. 16, supplement 1, pp. S213โS220, 1992.
[16] S. Yu, F. B. Ma, and B. Li, โOptimization of cracking furnaceoperation to lower energy consumption of ethylene plant,โEthylene Industry, vol. 22, no. 1, pp. 58โ59, 2010 (Chinese).
[17] Z. L. Wang, D. U. Wenli, and F. Qian, โIntelligent control tech-nique for ethylene plant,โ Chemical Industry and EngineeringProgress, vol. 25, no. 12, pp. 1454โ1456, 2006.
[18] Y.Han, Z. Geng, andQ. Liu, โEnergy efficiency evaluation basedon data envelopment analysis integrated analytic hierarchyprocess in ethylene production,โ Chinese Journal of ChemicalEngineering, vol. 22, no. 11, pp. 1279โ1284, 2014.
[19] J. C. Yang and Z. Wang, โAPC and RTO technology of ethylenecracker based on NIR,โ Automation in Petro-Chemical Industry,vol. 50, no. 4, pp. 1โ9, 2014.
Submit your manuscripts athttp://www.hindawi.com
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttp://www.hindawi.com
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
CombinatoricsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
International Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com
Volume 2014 Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Stochastic AnalysisInternational Journal of