CHP process evaluation, optimization and troubleshooting in OpenModelica

(1)

Master of Science Thesis

KTH School of Industrial Engineering and Management Energy Technology EGI_2017-0029 MSC EKV1184

Division of Heat & Power SE-100 44 STOCKHOLM

CHP process evaluation,

optimization and troubleshooting

in OpenModelica

(2)

Master of Science Thesis EGI_2017-0029 MSC EKV1184

CHP process evaluation, optimization and troubleshooting in OpenModelica

Hannes Wistrand

Approved

2017-05-30

Examiner

Miroslav Petrov - KTH/ITM/EGI

Supervisor Miroslav Petrov Commissioner Sigholm Konsult AB Contact person Fredrik Starfeldt

Abstract

(3)

SAMMANFATTNING

En viktig del i ett energisystem som trycker på hållbarhet men som också ska kunna leverera pålitlig energiförsörjning är optimering av kraftproduktion i alla led. Då mer än hälften av alla hushåll i Sverige nyttjar fjärrvärme som främst är ett resultat av förbränning av biobränslen och avfall men också av diverse fossila bränslen som alla är ger upphov till stora utsläpp är optimering av dessa energislag av stor vikt. Användandet av olja och kol, ofta i form av torv, är i synnerhet intressant att minska. De två bränsleformerna svarar tillsammans för nästan 10 % av

värmeproduktionen i Sverige och är i dagens fjärrvärmeproduktion nödvändiga reserver för att tillgodose en stabil energiförsörjning då de flesta reservpannor drivs på dessa bränslen. När driften av en panna driven på biobränslen måste stoppas sätts ofta reservbränslen in som utöver större miljöpåverkan också är mycket dyrare för fjärrvärmeproducenten.

För att minska den totala bränsleanvändningen i fjärrvärmeproduktion men också reducera de tillfällen en panna måste drivas på olja eller kol behövs en ständig process- och driftoptimering. Det finns ett flertal program som kan göra de båda. Men inom processoptimering är ofta programvarorna dyra i förhållande till potentiella besparingar. Därav är det av stort intresse att kunna utveckla programvaror som erbjuder en driftplanering som kan passa olika bränsletyper och efterfrågan på fjärrvärme men som samtidigt kan erbjuda en fingervisning om vad

värmeproduktionen beroende på diverse driftsvillkor bör vara då pannan ansen gå felfritt. Om denna möjlighet till jämförelse med en så nära felfri panna som möjligt finns att tillgå året runt finns möjligheten att i mätvärden se avvikelser som kan tyda på komponentfel som kan innebära oplanerade driftstopp. Detta samtidigt som en sådan modell kan ta ett stort antal driftparametrar i hänsyn för att vid olika tillfällen styra temperaturnivåer och

rökgassammansättningar m.m. i syfte att planera driften av pannan ifråga på bästa möjliga sätt. I denna avhandling har open source programmet OpenModelica använts för att konstruera en modell av en fastbränslepanna för just de ovan nämnda syftena att optimera och felsöka olika driftfall. Som fallstudie har Fastbränslepanna 1 på Hedemora Energi använts för att kunna dra slutsatser om programmets behov av inputs och dess förmåga att beräkna börvärden på

producerad ånga. En utvärdering av programvaran under utveckling och i användning utförs och resultat baserat på jämförelser mellan börvärden och uppmätta värden genomförs och diskuteras. Själva modellbyggandet i OpenModelica är mycket tidskrävande då det kostnadsfria programmet inte har någon välfungerande felsökning kopplat till beräkningsprocessen. Felsökning i

(4)

Dessa korrelationer användes därefter i komponenterna för att på så vis skapa en s.k. off-designmodell där vattentemperaturer och tryck istället för att anges beräknas med hjälp av korrelationer för det bättre skick pannan befann sig i under tidig höst.

Korrelationerna för värmeövergångstal och tryckfall utrycktes som funktioner av vattenflöde i de olika komponenterna och gav vid körning av styckvis komponenter önskade resultat.

Värmeövergångstalen ökar med last men konsekvent långsammare då lasten ligger över pannans designade maxeffekt på 10 MW. För numeriska skäl i OpenModelica kunde inte alla värmeväxlande komponenter köras i off utan detta beräkningsläge fick användas i de båda överhettarna samtidigt som resten av komponenterna balanserades manuellt. Då felen som ska gå att upptäcka förflyttar sig i pannan går de att se som resultat i just överhettarna.

För att validera off-designmodellen användes tre lastfall under tidig höst som skilde sig från de laster som pannan designats för. De gav goda resultat då tryckfall i modellen skilde sig mindre än

1‰ från uppmätta värden. För beräknade ångtemperaturer vid laster lägre än 7 MW låg största

felmarginal på 1.05%. För en last på 9.2 MW låg största felmarginal på ångtemperaturer på 2.5%, detta för ångtemperaturen ut ur överhettare 1.

För att se förslitning och sotbildning i pannan simulerades en lastperiod under en längre drift på hög effekt. Detta under mitten av februari. Här ger off-designmodellen de börvärden på

ångtemperaturer som pannan i gott skick ska kunna producera i jämförelse med uppmätta värden. Skillnader upptäcktes i form av att överhettare 2 i synnerhet konsekvent gav lägre temperaturer än vad som borde producerats vid samma lasttillfälle. Detta anses till stor del bero på just slitage och sotbildning men kan i viss mån tillskrivas ändrade bränslesammansättningar eller liknande. Detta märker därmed modellen av och indikerar att resultaten skiljer i för hög grad. Då driftstopp beroende på bränslepriser och efterfrågan på värme kan innebära stora förluster för värmeproducenter undersöktes modellens förmåga att upptäcka fel i pannan för att på så vis förhindra eller förkorta driftstoppet ifråga. Ett driftstopp som på Hedemora Energi skedde i början på December analyserades. Från det att modellen producerade resultat liknande de uppmätta genomfördes jämfördes resultaten löpande fram till driftstoppet. Ångtemperaturerna i båda överhettarna visade på stor divergens, i synnerhet i överhettare 2 där temperaturerna gick från en felmarginal på 2.6% till att succesivt öka närmre driftstoppet till en felmarginal på ca 15.7%. Även bränslesammansättning, rökgasåterföring och lufttemperaturer varierades för att visa på modellens flexibilitet i optimering. Vätskehalten i bränsletillförseln varierades för att visa på dess reducerande effekt på rökgastemperaturer, temperaturen på tillförs luft ökades för att höja densamma, flera olika åtgärder däribland de båda nämnda samt ökad rökgasåterföring prövades för att nå temperaturer mer skonsamma för de olika komponenterna i pannan.

(5)

-i-

List of Figures

Figure 1: A black box model of a boiler layout showing in- and outputs. ... 2

Figure 2: The tree menu showing the utilized packages ... 3

Figure 3: A simple heat exchanger tested with inputs and outputs ... 4

Figure 4: The water flow kept for all models of the boiler. (Bogart) ... 20

Figure 5: The flue gas flow in the first designed system. (Bogart) ... 20

Figure 6: The first layout tested in the design process... 21

Figure 7: The second layout tested in the design process ... 22

Figure 8: The load during 2015 for boiler 1 at Hedemora Energi ... 23

Figure 9: The load period containing the “design points” ... 23

Figure 10: The UA values of the economizer as functions of water mass flow ... 25

Figure 11: The UA values of superheater1 as functions of steam flow ... 26

Figure 12: The UA values of superheater2 as functions of steam flow... 27

Figure 13: The pressure drop in the economizer as function of water flow ... 28

Figure 14: The pressure drops in super heaters 1 and 2 as functions of steam flow ... 28

Figure 15: The first running off-design layout ... 30

Figure 16: The water and steam temperatures exiting each component at 20:00, the 25th of September ... 33

Figure 17: The water and steam temperatures exiting each component at 03:00, 28th of September ... 34

Figure 18: The water and steam temperatures exiting each component at 02:00 the 19th of October ... 35

Figure 19: The steam temperature exiting super heater 1 ... 36

Figure 20: The steam temperatures exiting super heater 2 ... 37

Figure 21: The load period preceding and during the downtime in early December ... 38

Figure 22:The steam temperatures exiting superheater 1 at even intervals, 4 days preceding downtime ... 39

Figure 23: The steam temperatures exiting superheater 2 at even intervals, 4 days preceding downtime ... 40

Figure 24: Temperature of flue gas entering each component, at unchanged input conditions ... 41

Figure 25: The flue gas temperature entering each component, as 50 g of water is added per second ... 42

Figure 26: The flue gas temperatures entering each component, using doubled recirculation ... 43

Figure 27: The flue gas temperatures entering each component, using increased air input temperature ... 44

(8)

-iv-

List of Tables

Table 1: The utilized media-dependent functions with in- and outputs ... 7

Table 2: Connector types and their transported variables ... 8

Table 3: Constants used for calculating the LHV of fuel (Wester, 2012) ... 18

Table 4: The necessary, assumed for, input parameters ... 19

Table 5: The “design points” chosen for the model ... 24

Table 6: The parameters varied to match measured values. ... 25

Table 7: The parameters supplied to the off-design model ... 29

Table 8: The implemented components and parameters for the final model ... 31

Table 9: The hours and loads used for validating the off-design model ... 33

(9)

-v-

Nomenclature

Parameter Abbreviation Unit

Water/steam temperature Water/steam enthalpy

Specific heat capacity cp /

Water/steam mass flow /

Pressure P Bar

Flue gas Temperature Flue gas enthalpy

Concentration -

Heat transfer coefficient UA kW/K

Logarithmic mean temperature difference LMTD K

Heat capacity rate C kJ/sK

Number of transfer units NTU -

Efficiency -

Molar mass M kg/kmol

(10)

-vi-

ACKNOWLEDGMENTS

This thesis was written as an assignment for Sigholm Konsult AB during the fall of 2016 as a Master’s Thesis at KTH’s department of energy technology.

I would like to thank Fredrik Starfeldt at Sigholm who as my supervisor at the company gave the necessary guidance and instruction for me to finish this thesis. Also, I would like to thank Niclas Sigholm for the opportunity to join the company during the fall and making the process of travelling to and working in Västerås much easier. Many thanks go to the staff at Sigholm Konsult for a nice social stay at the company.

I also want to thank Miroslav Petrov as my supervisor at KTH for giving me the inspiration and free reigns for producing the model presented in this thesis.

(11)

-1-

1 INTRODUCTION

District heating is largely utilized in Swedish households where about 50 % have access to it (Fjärrvärme). The production of district heating and steam generated electricity are both largely dependent on biofuels and waste, while there is still some fossil fuel in use for the same purpose today. The fossil fuel used for this purpose is mostly oil and peat. For heating purposes the fossil fuel usage in 2013 reached a level of about 70 000 TJ while the biofuels produced about 103224 TJ of energy (IEA). As both fossil fuels and waste contribute largely to the emissions in the heat and power sector the need for optimization and process troubleshooting is quite large and dependent on reliable analysis.

An integral part of this is the modelling process inherent in testing any thermodynamic process. Many programs are used for this exact purpose. Most of which work with a user interface in which one can select readymade program packages graphically in order to simplify the programming process. Often a flow diagram of the simulated power plant is used as a template for programing. This often implies certain difficulties in that the software packages are difficult to manipulate or rearrange in order to best fit a case study, given the variance of CHP plants and their respective constructions. An open source program such as OpenModelica can address these issues of conformity better than many others while offering a simple flow chart layout while it remains free for any user. Several teams have created standards of programming the most common components of heat and power plants so as to provide the groundwork for simulating a plant (Baligh EL Hefni, 2008). To further address the performances, troubleshooting and finally optimizing heat and power plants one may wish to manipulate certain components and implement code that can help in addressing where a certain loss may occur etc. As OpenModelica offers this possibility while it can be utilized with several relevant open source libraries it could prove to be a useful resource for anyone wishing to increase fuel efficiency in several heat and power applications using any fuel (Baligh El Hefni, 2011).

To validate the OpenModelica library created in this thesis, boiler 1 at Hedemora Energi is used as a case study and subsequently modelled. It is used to design components which are then implemented in a so called off-design model, to produce results similar to those measured while they may, if differences appear, provide clues to what different running conditions may have as consequences for the heat and power production, the condition of the components and what can be done differently.

(12)

-2- 1.1 Aims and Objectives

The overall objective of this thesis is to develop a steady state model of a biomass fuelled boiler in

OpenModelica (OM) and determine its versatility, stability and overall performance through

implementation in a case study.

More specifically the model is to be developed through building a component library consisting of conventional boiler components which are, under some limitations, able to be connected in cascade arrangements to match any boiler size or layout. This will make component troubleshooting easier as comparisons for every load can be produced.

The model developed is meant to deliver ideal results with low losses and is to be compared to the measured data of the case study used. The comparison should provide conclusions as to how well the certain components chosen work and what eventual losses may be due to during different hourly loads. The most desirable use of this will be to run the model in comparison with measured data in order to see tendencies that later may give rise to forced downtime of the boiler.

Lastly, the aim of this thesis is to evaluate if and how OM and the developed component library can be used in a commercial application. The expected outcome should be a working model used for evaluating hourly mean values of heat production and troubleshooting boiler components. The models ability to do this will be evaluated with respect to a viable margin of error.

The simple outline of how the boiler is to be modelled is displayed in a simple schematic below, showing the necessary inputs and the boiler in which one can arrange components as in any real arrangement.

(13)

-3- 1.2 Methodology

To meet the described objectives of this thesis a methodology outline is presented here. It is used throughout the work conducted here for the purpose of discussing acquired results and drawing conclusions based on them and the objectives set.

1.2.1 Model layout

To create a modelling standard for boilers one needs to construct a library of conventional components that can be arranged in a specific order to match any layout desired. After a certain layout is built it should be run using input values from a case study while output results are compared to those of the same. This should give the difference between an ideal boiler of specified layout and that which is analyzed.

OM is inherently built using so called packages which simply put are folders containing everything

a model needs to utilize in order to run. More specifically this encompasses all code one needs to write. When saving a collection of code, an entire project or the like, this should be in the form of a package. This package can thereby be loaded by the program and everything the package contains can subsequently be used. The main package for loading in OM is simply called Boiler in this thesis. It contains all layouts one wishes to create, a subfolder/package containing the component library, one containing the connectors and one containing functions. The natural use of the package system is through a tree view menu displayed in a simplified manner in figure 3.

Figure 2: The tree menu showing the utilized packages

(14)

-4-

1.2.2 Component testing

Components are to be built taking their respective functions into account, this with regards to thermodynamics, mass balances, energy balances etc. and then tested separately with system boundaries at either side of the component. The boundaries are here inputs and outputs for both media i.e. steam/water and flue gas. This build up is made to assure that the component works as it should, transporting the correct variables to the outputs. A simple heat exchanger is displayed in

figure 2 to give an intuitive picture of how the components are tested separately.

Figure 3: A simple heat exchanger tested with inputs and outputs

1.2.3 Component dimensioning and off-design modelling

When a library is readymade and components tested, they are to be arranged as in the layout of the case study considered here. Whereby the comprised model is to be run using measured values from a few load periods that are as even as possible and of different magnitudes. Varying the fuel input until the flue gas temperature exiting the model matches that of the measured values should then give the desired heat balances. This will assure, even though point-specific temperature measurements may differ, that the heat absorbed in each component is correct given the specific layout of the model. As the boiler considered here is dormant during most of the summer during which maintenance and cleaning usually is made two distinct periods are utilized. Early fall is used for design conditions, i.e. using known parameters in order to calculate UA values of different components at different loads. The UA values are then as close to ideal for the case study as possible during the year. This leading to correlative expressions for the UA values as functions of water/steam mass flow.

(15)

-5-

for the components between the feed water pump and the furnace. With the only exception of the De-superheater in which a certain temperature must be expressed for which the steam is not allowed to surpass.

The layout designed for should be chosen with some care with regards to the calculating nature of

OM and not only with regards to the actual layout of the plant. Most apparent when planning

approximation should be the downcomers utilizing natural circulation while providing steam to the deaerator. As these are pipes laid out in a, for the examined case study, unknown arrangement throughout the boiler walls the exact placement in a cascade arrangement of components is impossible to model correctly. Therefor the placements of these are experimentally tested in order to provide the best possible results. More specifically the flow of flue gas through the downcomers is drawn in different orders with regards to the other heat exchanging components of the boiler. These expressions for heat transfer rates are then used in off design conditions where heat balances are conducted via the NTU (number of transfer units) method in order to calculate in and out temperatures of the components at a given mass flow rate of water enters the boiler. Using this off-design model, the mass flow of water and fuel are given and as result the load and temperatures inside the boiler are calculated.

1.2.4 Analysis

The off-design model described is to be used to analyze a period which should display more fouling and wear while the early fall, designed for, should display results close to those of the developed model. Hopefully the off-design model may hint at where different errors in the boiler may occur and what they can be due to. Furthermore, the off-design model can provide the necessary tools to figure out where different optima may occur. At which amount of air in excess, at which amount of flue gas that is recirculated to the furnace, at which water content in the fuel a desired temperature is reached etc. as the maximum temperatures, are not always desirable. The different flue gas temperature limitation methods such as water content in the fuel and flue gas recirculation are to be varied and their consequences evaluated.

Furthermore, what should be run in the model to test its versatility and finalizing the evaluation of its actual usefulness the model should be used in order to find discrepancies that in turn might lead to downtime for Boiler 1 at Hedemora Energi. This as one of the larger costs for CHP plants and in industrial activities in general is the downtime needed for fixing an unexpected fault, relevant to this thesis, in a boiler component (Nepal, 2004).

1.3 Demarcation

There are several demarcations in this thesis. This as there are many variables and parameters interacting in unforeseeable ways. These demarcations are listed below

(16)

-6-

 Only 7 fuel components are considered in the combustion section. These include carbon, hydrogen, nitrogen, oxygen, Sulphur, water and ash. This leads to 6 components in the flue gas namely carbon dioxide, nitrogen, Sulphur dioxide, oxygen, water and ash.  The system the developed components encompass start at the feed water pump for the

water side of the system. Thereafter ending in a steam output after the last implemented super heater. The flue gas originates in the furnace, utilizing fuel and air inputs and ends in a flue gas output after the economizer. The flue gas can partly be recirculated to the furnace. This implies that no components utilizing the produced steam are considered. Neither are components handling the flue gas outside of the boiler as stacks or the like.  No fluid mechanics is considered to calculate pressure drops or convective heat transfer

coefficients.

 The model is based on steady state calculations. Constant mass flow of both cross-flow media.

 No UA value is considered in the down-comers. This as the governing equations don’t apply when phase change is under way.

 Flexibility regarding input parameters is not considered. No code is written to handle several input options, i.e. more than one value for each parameter.

(17)

-7-

2 MODELLING STRATEGY

Working with Modelica to model heat and power systems has been done in the past. However, libraries and packages developed are seldom open-access or compatible with the open source

OpenModelica updates but better with non-open source programs such as Dymola. To add to this,

they have widely differing operating parameters, inputs and outputs. They are therefore not very flexible in adapting to whatever measurements are available from case studies. This while they often display redundant information in the same application (Quoilin et al., 2014).

To maintain a desired degree of flexibility and end purpose of calculation a new boiler-library is developed in this thesis.

A prerequisite for all components is that required inputs must be defined to solve for desired variables. The model developed in this work utilizes the greater reliability of water and steam temperatures as to differ from flue gas temperatures and subsequently calculates the flue gas temperatures as one of their respective outputs (Trinks et al., 2004).

2.1 Functions

Common for most components is the function package. This is a folder containing the different functions defining fluid properties such as enthalpy, temperature, pressure etc. called for in every component. Functions utilized are shown in table 1 listed by what fluid they handle, what their respective inputs and outputs are. The property functions of water and steam are inherit in OM and are thereby not necessary to specify. The property functions of flue gas are specified using polynomial approximations that are temperature dependent. These functions and their input coefficients, , are therefore displayed in appendix 1.

Function Output Fluid Media Required Inputs

h/h’ Water/Steam P, T

h’’ Steam P

T’’ Steam P, h

h Flue gas c, T

Specific flue gas components 1_{, T}

T Flue gas h, c

cp Steam h, p

Table 1: The utilized media-dependent functions with in- and outputs

(18)

-8- 2.2 Connectors

Connectors in Modelica are used as nodes by which one can attach models to each other and thereby transport variables between them (Tiller). This is done graphically in the diagram window in OM and represent the flow of chosen media. In this thesis, there are 5 types of connectors utilized. In these the mass flows are defined as flow variables, meaning that a certain direction is specified other than stating a certain value of the specific point the connector represents which is the way in which the remaining variables are handled. In table 2 the connector types, and their respective variables are stated.

Connector Variables Water/Steam , , , Flue gas , , , , Air , Ash Fuel , , , ,

Table 2: Connector types and their transported variables

The connectors listed in table 2 can all be used as long as the transported variables are utilized or further transported in an attached component. To note here is that in the model developed to fit the case study given for this thesis the connectors handling water, steam and flue gas are the only ones acting as both in- and outlet ports.

2.3 Components and their design

The main parts of the model layout are the components. The actual library. The components are built as so called models in Modelica. That means that they incorporate the main parts of code utilized. The parts of a system that manipulate and utilize transported variables and node variables. The library developed in this work consist of different categories of components such as heat exchanging components, a furnace, flow splitters etc. that together can be arranged to fit common boiler types. These components will inherently, due to Modelica’s programming nature, display calculated variables The different components are further explained in this chapter.

2.3.1 Feed water pump

(19)

-9-

design of the feed water pump follows that of a conventional sketch of a pump while the placement of the connector implies the flow direction.

The parameters supplied here are the mass flow of water, the pressure supplied by the feed water pump and the water temperature. The enthalpy of the water is then calculated. These are then transported as variables from the port to the next component in the chosen order.

2.3.2 Economizer, Super heater and Deaerator

The heat exchanging components that can be arranged in cascades and utilized in more than singles are the super heaters and the economizer while the deaerator, which can can be arranged in the same way, is usually a single component. They are components with connectors of two common types as both inlets and outlets. These are in the form of water/steam ports and two flue gas ports. Outlets are denoted as port2 and inlets as port1. They are to handle the states of water in and out as known parameters and calculate the output flue gas temperature and enthalpy. This will make the flue gas mass flow and temperature the governing factors for knowing if the calculation process crashes or not. This as the heat in the flue gas, as calculated in equation 2, must be greater than that of required energy to heat the water as in equation 1. This is the main foolproof system, inherit in the coding of Modelica. If the heat required surmounts that which is supplied by the flue gas, the logarithmic mean temperature difference (LMTD) will depend on a negative argument resulting in no values for input in equation 3 leading to a calculation error and subsequently halting all other calculations.

For the deaerator and its downcomers no LMTD is solved for which makes the component rely on another function stating that if the fluegas temperatures are lower than the water temperatures on the corresponding sides the process is set to crash.

The first of the above-mentioned security measures is more specifically structured through calculating the enthalpy of flue gas exiting the component in question and compare it with an iterated variable starting at zero. And if the enthalpy is smaller than that which is iterated the entire process is set to crash. The first safety measure is completed using a second looking at the value of the heat that is exchanged. This could without measures turn into a negative value, were the steam/water to heat the flue gas, and is therefore used as a condition. Namely, if the heat is lower than zero, all calculation is halted.

As to the arrangement of these components one should as conventions has it connect the economizer to the feed water pump. It’s input of flue gas should connect to the flue gas output of the last super heater utilized, i.e. that which handles the lowest temperatures or the output from the downcomers belonging to the deaerator.

(20)

-10-

The economizer is similar to the superheater but created so as to differ from it if any off-design differences should appear. This graphical difference will also help in clarity when using the interface.

The parameters supplied to the economizer are the same as for the superheaters.

The deaerator differs from the two heat exchanger types as the heat required is mostly latent heat i.e. the heat required to vaporize the water provided. The known pressure will therefore be the only thing needed to calculate output enthalpy of the water and flue gas. No parameters are thereby necessary to provide in this component.

2.3.3 Flue gas splitter

The flue gas splitter is utilized to specify the percentage of flue gas that takes a certain route through the system one wishes to model. A concrete use for it is to specify the amount of flue gas that is ventilated to the stack and how much that is recirculated to the furnace for cooling purposes and reduction (Baukal, 2003). The only thing specified is the percentage of mass flow that is directed to port2.

2.3.4 Flow splitter and De-superheater

As feed water can be utilized for two purposes. Namely, feeding water into the main heating process starting in the economizer, and cooling the steam between the super heaters with the De-superheater, sometimes called the de-superheater (Ghaffari et al., 2007). To achieve this, a flow splitter is used here. If a certain inflow of water is calculated for or specified as a parameter in a component connected to the one of its ports, the other mass flow is specified as the remainder of the feed water. The De-superheater utilizes a set temperature value as a limit between the super heaters. The enthalpy at that temperature is solved for and subsequently the necessary mass flow of feed water needed to reach the limit temperature is solved for and drawn from the water splitter. These two components are necessary to utilize simultaneously and connected in the specified manner. The flow splitter for water is designed in the same manner as for the one of flue gas in figure 7, differing only in color and is for consistency reasons blue. The De-superheater is displayed in figure 8.

The De-superheater in figure 7 has 3 connectors handling the water and steam flows. The top connector handles steam into the valve, the bottom one handles steam output and the one on the left side of the valve handles the feed water flow that is diverted from the flow splitter.

2.3.5 Furnace

The furnace is integral to every boiler and no layout should be without one less one uses a gas turbine cycle or the like. However, in this thesis the furnace exclusively handles solid fuel.

(21)

-11-

Parameters specified in the furnace are the air excess factor varied to match the measured O2 surpluses and the ash removed. The adiabatic process calculates the air flow into the furnace The necessary inputs are further described in section 2.3.7.

To note in figure is that the furnace has 5 connectors. The two red ones are the flue gas exit at the right side, where the flue gas produce enters the boiler, and the flue gas recirculation entering the furnace on its left side. The brown connector represents the fuel input, the light blue the air input and the grey the ash output.

2.3.6 Bottom ash conveyer

If one wishes to remove the flue gas ash content in the boiler one can do so specifying the percentage of ash removed either in the furnace or in new components one could implement called bottom ash conveyers. They are programmed in such a way that using known percentages of ash removed they calculate new flue gas mass flow and concentration.

2.3.7 Inputs and outputs

As mentioned in section 2.2 there are different connectors corresponding to different transported variables. The inputs present in the library are those of fuel and air. Handling water is the feed water pump mentioned in section 2.3.1 while flue gas is a result of inputs being sent to the furnace. These inputs correspond to fuel and air. They combine to form flue gas which after passing all utilized components is discarded in the flue gas output. The steam produced in the last superheater is likewise sent to a steam output. These two signify the system boundaries. The different inputs and outputs are designed as arrows much like in figure 3. The difference being the color coding. The colors of the in- and outputs correspond to their respective connectors with one exception. This being the water connectors that are of purple color.

The parameters specified in the fuel input are fuel composition, specific heat of the fuel, its temperature, and the mass flow of said fuel.

(22)

-12-

3 The Concept of Design and off-design

As described in section 1.2.3 model of boiler 1 is first to be designed, meaning having its dimensional and flow dependent variables calculated and expressed as the exact areas of the heat exchangers and flows that are in contact with their areas are unknown and prone to vary. Here follows a description of the designing process and what differs who the off-design model differs from that which is used for the dimensioning.

3.1 Designing and dimensioning

To create a model running in design mode the components need to be dimensioned correctly per a certain set of measured values that must be supplied. The dimensions in this thesis are expressed as heat transfer coefficients, UA-values, which in turn are dependent on surface area. If one can use the necessary heat balances and calculate the UA values as functions of water flow, which in turn is constant, the correlation should hold for that specific water flow and the current flue gas properties.

To get the dimensions described and heat balances that hold, flue gas properties need to be calculated while water and steam temperatures as well as pressures are continually supplied to the model in design mode. As a furnace that handles the combustion processes is applied to take fuel compositions into account the different flue gas temperatures will not match those measured. This also depends on the flow order of the flue gas which in turn depends on the programs capability to pass variables from component to component. However, as the UA-values are built to fit a certain condition and relationship between flue gas and water, the flue gas properties produced can be calculated as products of the specific fuel used in this thesis. This while the flue gas output temperature exiting the modelled boiler is matched to the temperature measured at the stack. This will subsequently supply the user with a known amount of heat transferred until the flue gas leaves the system.

The same procedure should be done for the pressure drops. They do not require additional equations and will only require making correlations with water flow.

Once the UA values and pressure drops all are correlated and made into functions of water flow the off-design model can be constructed.

3.2 Off-design

(23)

-13-

The end goal of this procedure is to have a model that can calculate the steam and flue gas temperatures as well as steam pressures as ideal values.

(24)

-14-

4 Thermodynamics

4.1 Heat exchanging

The heat exchanging components of the boiler only display one difference it being the physical state of water passing through. The economizer only handles liquid water while the super heaters handle steam. As the economizer handles liquid water it displays a much larger pressure drop which needs to be considered. This as pressure drop is dependent upon density. Specifically, the pressure drop increases as the density increases (Havtun, 2014). As this is considered the same thermodynamic equations can be implemented in these heat exchangers.

4.1.1 Design calculations

The components are to be utilized so as to calculate output properties of the flue gas. The properties of the flue gas stream that passed the component that is. This is done through a simple energy balance. Firstly the heat needed for the water flow is calculated as

(1)

And is set equal to the heat supplied by the flue gas

(2)

Which solves for and the dimension of these components are subsequently estimated using the heat exchanged in in crossflow exchange

∙ (3)

Where LMTD is the logarithmic mean temperature difference and calculated as

ln

(4) Equations 1-4 provide the necessary outputs needed to match the measured values of the case study. The margin of error should thereby provide grounds for conclusions regarding the management and maintenance of the boiler. As these components are arranged in a layout close to cascade arrangement and by necessity influence each other a common solution is solved for here. For the deaerator only equations 1 and 2 are utilized for heat balance purposes. This as the latent heat supplied won’t provide a significant temperature difference for the water and steam. This leads to the LMTD being redundant.

(25)

-15-

The flue gas temperature is calculated as well, this through running enthalpies for every temperature from zero to that of the one matching. This temperature is thereby displayed. The function is specified in table 1.

The heat transferred in the steam at the De-superheater (de-superheater) is calculated as

(5)

Which is the heat required to heat the cold feed water diverted from the flow splitter to the limit temperature set here. This heat is set equal to

(6)

Which is the heat that is required to be removed from the flow coming from the first super heater, , to reach the limit temperature. Equations 5 and 6 will thereby give the two mass flows.

4.1.2 Off design calculations

After calculating the UA values for the economizer and superheaters these are specified as functions of water flow. The components are thereby to calculate all water temperatures as well as the output flue gas temperatures. However, this may stipulate a problem in calculation as there are too many unknown temperatures which subsequently could provide an unknown number of solutions depending on the start values given in each component. This as the start values most likely should be changed in accordance with the boiler load etc. This is not desirable from a modelling perspective.

The off-design modelling is therefore initiated by using a different method of estimating heat and temperatures, namely the Number of transfer units-, NTU- method. This utilizes the so-called heat capacity rates of the different flow media. Heat capacity rates for flue gas are calculated as

∙ (7)

And utilizes the input temperatures and enthalpies for flue gas in every component. This while the heat capacity rates for steam are calculated through

(8)

The heat capacity rates are compared and the smallest of the two is after that defined as and the larger as . The ensuing step is that of calculating the largest amount of heat that can be transferred theoretically. This is defined as the cold media entering the heat exchanging component reaching the input temperature of the hot fluid entering the component, i.e. the largest temperature difference on both sides of the heat exchanger. This heat is calculated through

(9)

(26)

-16-

(10)

and is further utilized to estimate the efficiency through 1

1

(11) For super heater 1 which is arranged for cross flow media.

While the efficiency for super heater 2, in which there is parallel flow, is estimated through 1

1

(12)

Where is expressed as

(13)

The real heat transferred in the heat exchanger is thereby calculated as

(14) This heat is then set equal to the ones calculated in equations 1 and 2 in order to calculate the output enthalpies and subsequently their corresponding temperatures.

4.2 Combustion

As mentioned in section 2.3.7 the furnace is connected to two inputs. These handle fuel and air. The furnace takes composition and temperature of the fuel and air into consideration and thereby calculates the flue gas concentration through adiabatic combustion. Using molar weights and percentages one can calculate the amount of moles per kilogram of fuel through

/ (15)

(27)

-17-

The number of oxygen moles needed per element present in the fuel is shown in appendix 1. The required amount of surplus oxygen, more than what is present in the fuel, for the adiabatic combustion is subsequently calculated as mentioned above. This results in a known amount of nitrogen added assuming atmospheric air is supplied to the combustion process. This nitrogen is inert in an adiabatic process and the number of moles per kilogram of fuel is calculated as

3.76 (16)

Subsequently providing a substantial fraction of the flue gas mass.

As mentioned in section 2.3.5 there is often a surplus of oxygen in present to assure a more complete combustion. This is accounted for through adding the surplus to the known number of oxygen moles, , and subsequently doing the same for the number of nitrogen moles solved for in equation 7.

If one wishes to use flue gas recirculation the flue gas compositions will differ to the composition due to combustion. This as most of the fly ash is removed before the recirculation. In the model developed the flue gas recirculation is specified as a percentage of this composition and is used in the same manner as the oxygen surplus namely

(17)

The flue gas concentration, , does not differ much from the flue gas concentration resulting of pure combustion, , as the ash content often is close to negligible in mass concentration. To calculate the enthalpy of the flue gas exiting the furnace a heat balance is utilized. Firstly the energy released from the combustion can be estimated using a relation for lower heating values for solid fuels which is specified as

(18)

Where the constant k for each element in the fuel is displayed in table 3.

Fuel element Value of k

Carbon 34.1

Hydrogen 101.8

Nitrogen 6.3

Sulphur 19.1

(28)

-18-

Water -2.442 Table 3: Constants used for calculating the LHV of fuel (Wester, 2012)

The LHV is utilized in heat balance as

(19)

Secondly preheated air is taken into consideration as well as recirculated flue gas. This energy is estimated as

(20)

Where the enthalpies are calculated using a polynomial handling all gases in the model. This polynomial is found in appendix 1. Equation 11 gives and . If the fuel added is preheated as well the heat must be accounted for here. It is estimated as

(21)

Where the reference temperature is the same as for gases, namely 0 .

A certain amount of heat is also needed in order to vaporize the water content of the fuel. This is calculated as

(22) Where is the heat of vaporization at atmospheric pressure.

The heat generated is then added to give a flue gas energy content. This is done through the balance

(23)

Using equation 14 and the calculated for flue gas power the output enthalpy can be solved for explicitly through using equation 11

(24)

Which in turn can be used to find its matching temperature. If the temperature is known all gas enthalpies are calculated as

(25) Where the concentrations are derived from equation 17 and the component specific enthalpies,

(29)

-19-

5 Solving under design conditions

To run a model under design conditions one needs a necessary amount of input data for the model to provide the required output. A model under design conditions is made as described in chapter 2 and 3 meaning that several components are made to fit any layout. Designing in this regard is thereby using these components, necessary input, and output data to calculate the UA- values and pressure drops as relations depending on the boiler load.

Common for the designing processes is that the temperatures and pressures of the steam and water are given and their enthalpies solved for. These are utilized in the heat balance in equation 1, set equal to equation 2 and should therefor provide the temperature and thereby enthalpy of the output flue gas for each component. This works through first calculating the adiabatic flame temperature of the combustion process, accounted for in section 3.2, which is used as flue gas temperature input in the first heat exchanging component which in turn depends on the cascade order described in section 1.3.3. In that specific component, the flue gas output temperature is calculated and is thereby, as the cascade order suggests, utilized as input temperature for the second component in that layout.

For the different layouts designed for, the fuel input is varied until the flue gas entering and exiting the economizer-area reaches the same temperature as the one that is measured and

supplied by Hedemora Energi. The necessary parameters that have not been measured and supplied are presented in table 4 and used throughout the modelling as keeping them constant would simplify the modelling process.

Parameter Value Unit

Flue gas recirculation 102 _{% of total flow}

Oxygen in air 213 _{% of total}

Nitrogen in air 792 _{% of total}

Air temperature 36

Ash removed 80 % of total

Table 4: The necessary, assumed for, input parameters

To note here is that the fractions of nitrogen and oxygen are modified from what the source tells us. This as no functions of enthalpy are accounted for when it comes to the other inert gases that normal air contains. As nitrogen is kept inert in the adiabatic process mentioned in section 3.2, it will in this thesis represent all inert gases of air while oxygen will represent its fraction given. To note also is that there is no consensus on commonly used fractions of recirculated flue gas. Therefor an arbitrary amount is used here which is used and tested for by other users.

The layout of boiler 1 at Hedemora Enregi is mimicked as much as possible from looking at the blueprints and control room interface. This provides the necessary knowledge to model the two media in counterflow correctly. Or as mentioned in section 1.3.3 in the right order while observing

(30)

-20-

the direction of water flow while the most suitable modelling order of flue gas flow is to be discovered during the modelling process.

The water flow is drawn in the same manner regardless of flue gas flow as this is known throughout the boiler and is used as a defining attribute of the same. The order of water flow is shown in figure 4.

Figure 4: The water flow kept for all models of the boiler. (Bogart)

As can be seen in figure 4, the only deviation from the main water flow is the water flowing to the De-superheater used to regulate steam temperature between the two superheaters. This usually to keep the output temperature of steam to a certain desired value (Ghaffari et al., 2007). This value is then supplied to the De-superheater as a set temperature and the flow from it to the superheated steam is then calculated using equation 5.

Secondly, the order of flue gas flow was chronologically, exiting the furnace, flowing through the downcomers where it supplied the heat necessary to evaporate the steady state flow of water. After which the flue gas entered the two superheaters supplying heat to the produced steam. The flue gas is lastly drawn to the economizer, where it preheats the water entering the deaerator and its downcomers mentioned above. This order and the percental amount of flue gas entering each component is drawn in the figure 5 below.

(31)

-21-

To note here is that the recirculated flue gas is drawn back to the furnace to achieve its purpose of decreasing in the combustion process (Baukal, 2003).

The first layout tested and run so as to produce the desired UA-values needed for creating an off-design model is displayed in figure 6.

Figure 6: The first layout tested in the design process

(32)

-22-

To analyse all data a new layout needs to be used so as to avoid the above stated problem with the layout in figure 6. A new layout tested entails only one change. This layout is displayed in figure 7.

Figure 7: The second layout tested in the design process

(33)

-23- 5.1 Loads used for design

As mentioned in chapter 1.3.3. the points for designing the model should consist of load periods close to early fall when the boiler is run after being dormant during most of the summer. The points for design should be chosen at periods which are not too unstable while still representing loads evenly distributed between minimum and maximum values. In figure 8 a plot of the boiler load is displayed.

Figure 8: The load during 2015 for boiler 1 at Hedemora Energi

The points used for designing purposes, furhter on called design points, are sought for in the area between 6400 and 7850 which are numbers correspondning to the hours of the year 2015. This period is displayed in figure 9.

Figure 9: The load period containing the “design points”

(34)

-24- Point Load [MW] Corresponding hour 6429 3.122 17:00, 25/9 6434 4.094 22:00, 25/9 6918 8.419 02:00, 16/10 7242 12.276 13:00, 29/10

Table 5: The “design points” chosen for the model

(35)

-25- 5.2 Dimensioning

As mentioned in chapter 4 the layouts tested are of great importance in evaluating the model as there is no clear arrangement for the deaerators downcomers. However, using the layout showed in figure 7 UA values for each component can be calculated to produce results identical to the measured values for the different states of water, steam and flue gas at each side of the economizer. The parameters varied and their dependent result that is to be matched to measured values finally giving the geometrical correlations are presented in table 6 in chronological order.

Varied parameter Variable to be matched

Air surplus Oxygen in flue gas

Fuel mass flow Flue gas output temperature

Water temperature from economizer Flue gas temperature into economizer Table 6: The parameters varied to match measured values.

As the parameters and variables above are matched to fit the measured values of said variables the model is balanced and should provide the UA-values sought for. Using the design points presented in table 5 their respective component-specific UA-values are plotted as functions of their respective water flows. In figure 10 the UA-correlation for the economizer in a relatively clean state is plotted.

Figure 10: The UA values of the economizer as functions of water mass flow

Noticeable in figure 10 is the close to linear line of UA values increasing with water flow. Two lines are distinguished, one for water flows beneath 1.5 kg/s and the other for water flows above the same. The two functions are expressed as

6.296 6.3479 (12)

For water flows under 1.5 kg/s and 0 5 10 15 20 25 30 0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 UA-value

(36)

-26-

3.0111 11.541 (12)

And are both to be implemented in the economizer belonging to the off-design model. The same procedure is repeated for both super heaters designed for. The correlation of UA values of the first is plotted in figure 11 and the one for the super heater 2 in figure 12 below.

Figure 11: The UA values of superheater1 as functions of steam flow

In figure 11 it is as in figure 10 appropriate to approximate the UA-values with two separate linear functions. The first for water flows under 2.6 kg/s and the second for flows above that value. The first is expressed as

0.5394 0.1496 (20)

And the second as

0.2312 0.9811 (20)

And subsequently utilized in the off-design version of super heater 1.

In figure 12 the exact procedure is repeated and its results shown for super heater 2. The best mathematical approximation for UA-values in super heater 2 is a third-degree function expressed as

0.0444 0.5156 1.9615 1.2948 (20)

Which is, as a safety precaution, extrapolated to show that within the span of the water flow used for Boiler 1 in 2015 the function won’t diverge to give extreme values. This function is implemented in the super heater of the off-design model as well.

0 0,5 1 1,5 2 2,5 0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 UA-value

(37)

-27-

Figure 12: The UA values of superheater2 as functions of steam flow

To add to figures 10 to 12 the pressure losses for water and steam are plotted similarly.

The pressure loss inside the economizer is shown in figure 13. It is plotted as a function of the components water flow.

The pressure loss in figure 13 is as in figure 12 approximated with an extrapolated third-degree function expressed as

0.0108 0.371 0.1866 12.619 (20)

The pressure drop for the two super heaters are plotted in figure 14. As the initial assumption for pressure drop here is that it is evenly distributed over both super heaters, i.e. from the deaerator to the steam output, the only difference is the steam flow resulting in the pressure drop for super heater 2 being slightly offset.

In figure 14 the line slightly offset to the left represents the pressure drop for the first super heater and is approximated with a second-degree function expressed as

0.0544 0.1243 0.119 (20)

while the one slightly offset to the right represents the pressure drop in super heater 2 and is approximated similarly through

0.071 0.0071 0.0166 (20)

giving pressure drops of the same magnitude but at different mass flows. -1,5 -1 -0,5 0 0,5 1 1,5 0 1 2 3 4 5 6 UA_value

(38)

-28-

Figure 13: The pressure drop in the economizer as function of water flow

Figure 14: The pressure drops in super heaters 1 and 2 as functions of steam flow

(39)

-29-

6 Solving under off-design conditions

When the expressions of UA value and water flow correlations are readymade they are

implemented in the off-design components. The economizers and super heaters. They are first to utilize the NTU method described in 3.1.2 and will subsequently calculate the different heat balances that solves the entirety of the model. This depending on the water flow through each component.

The first validation should then be carried out using the same loads as designed with. These should provide the grounds used for making a feasible error margin which can then be used along with several other loads to validate the model.

When switching to off-design conditions the numerical methods for solving changes. Some software are given the capacity for making the necessary assumptions depending on input parameters given. OM however does not offer this flexibility, thus forcing the changing of solution method in the components utilized. This results in a numerical evaluation of the methods accompanied with additional design factors that must be implemented further so as to have a model that requires as few inputs as possible.

6.1 Switching methods

To first create the off-design model one must decide which parameters are to be changed into solved for variables and which are to be used as inputs as in the designing of a model. Table 7 shows which input parameters are to be supplemented to the first working off-design model in order to give the desired solutions.

Component Supplied input value

Feed water pump , ,

Flue gas input De-superheater Flue gas output

Table 7: The parameters supplied to the off-design model

Introduced for the first time in table 7 is the component called flue gas input. This component is implemented for off-design calculations as the furnace, and its inherent equations in chapter 3.2, is only able to run if fuel mass flow is given. While it is possible to adjust the furnace to the input parameters chosen and displayed in table 7, it will provide the model with too many parameters making the system, solved by OM, overdetermined. The flue gas input is therefore to utilize the results of the furnace outputs to create a feasible simplification of the combustion process.

(40)

-30-

gas entering them as can be seen in equation 6. To come to terms with this problem and to find the necessary method for solving the off-design model the dependency of water temperature exiting the economizer and the flue gas temperature entering it needs to be further specified to avoid an infinite amount of solutions. However, inconsistencies also arise when trying to fix a temperature inside the boiler making the system over determined yet again. Therefore, the flue gas temperature exiting the economizer is specified along with switching back to the LMTD method for solving the heat balance here. As the flue gas temperature exiting the economizer and the water temperature entering it both are specified a solution of both medias input temperatures will be found. This results in the deaerator finding only one solution of necessary input heat, i.e. heat of the flue gas exiting super heater 1. This means that the numerical solution for the super heaters needs to be able to solve for an output temperature being given. This implies that the two super heaters should solve for either the flue gas temperature entering the boiler, out of what would be a furnace, or the mass flow of the same. As the adiabatic flame temperature in

combustion is dependent on the excess air the flue gas mass flow is chosen to be solved for here. In the flue gas input the different concentrations of component gasses are calculated as functions of air-excess factors given in the designing process. This is possible as the fuel concentration in table 3 is kept constant just like the flue gas recirculation and air concentration are, both presented in table 4. The air excess factor is however also expressed as a function of the approximate O2 surplus measured. This so as to supply only that in the flue gas input.

The resulting boiler layout following the restrictions and simplifications mentioned here is displayed in figure 15.

Figure 15: The first running off-design layout

(41)

-31-

While the layout in figure 15 provides results like those of the model in design mode the, this at a maximum error margin of 1.38 %, a large part of the models purpose goes amiss. This as no different fuel compositions or air excess factors can be varied and analyzed. As this is an integral part of the thesis while being the part of the model that provides most groundwork in lowering the fuel usage one more layout is built to encompass the necessary components.

6.1.1 Finalizing the off-design layout

In order to optimize and troubleshoot one would rather have the specific flue gas flow corresponding to chosen fuel concentration, mass flow and air excess rather than solving for a necessary flue gas flow. This using flue gas temperatures on either side of the boiler. In order to utilize the furnace and its accompanying inputs and outputs the solution method must be varied. As the furnace and its flue gas recirculation is implemented in the off-design model the inconsistencies described in 5.1 arise yet again as the system becomes over determined. Yet a change in numerical solution must be made in order to fit the new system of parameters applied. These parameters are described in table 8.

Component Supplied input value

Furnace ,

Fuel input , , , ,

Air input

Ash output -

Flue gas flow splitter %

Table 8: The implemented components and parameters for the final model

(42)

-32-

(43)

-33- 6.2 Validating the model

In viewing results a validating process is conducted here. This to provide grounds for the inherent error margins in the model. This is utilized to enable one to draw conclusions regarding other results and their relative error margins. Three points are chosen for different loads and their respective results are shown here.

The points chosen are presented in table 9 below.

Point Load [MW] Corresponding hour 6432 4.92 20:00, 25/9 6487 6.14 03:00, 28/9 6990 9.15 02:00, 19/10

Table 9: The hours and loads used for validating the off-design model

Figure 16 shows the water temperatures exiting each component. This for measured values and calculated for in the off-design model. The time for which this is done corresponds to 20:00, 25th of September.

Figure 16: The water and steam temperatures exiting each component at 20:00, the 25th of September

The offsets are, at this hour, so small that the model values completely obscure those measured. The largest offset corresponds to the steam temperature exiting super heater 2 and amounts to 0.23%.

In figure 17 the water temperatures exiting each component is shown, this at the point corresponding to 03:00, 28th_{of September.} 100 150 200 250 300 350 400 450

Pump Eco Dea SUP1 Valve SUP2

Temperature

Component

(44)

-34-

Figure 17: The water and steam temperatures exiting each component at 03:00, 28th of September

As can be seen in figure 17 the trends of each temperature increase in the model is like that of the measured values. The largest temperature offset here is the steam exiting the first super heater. It amounts to 1.05%. This while the error in steam temperature exiting super heater 2 only reaches 0.2 ‰. 100 150 200 250 300 350 400 450

Temperature

Component

(45)

-35-

In figure 18 the exiting water temperatures of the boiler components are shown at the time 21:00, 11th_November.

Figure 18: The water and steam temperatures exiting each component at 02:00 the 19th of October

The largest offset is seen at super heater 1 and amounts to 2.5 % second is the one at super heater 2 at 1.28%. As these are the largest errors the validating process these are assumed to be the largest occurring while the boiler runs at close to ideal conditions.

100 150 200 250 300 350 400 450

Temperature

Component

(46)

-36-

7 Results

When the off-design model can be run the different outputs of other load periods during 2015 can be analyzed and plotted in order to provide conclusions regarding the models versatility.

7.1 Fouling

In validating the model some offsets were seen. Most noticeable are the two steam temperatures exiting both super heaters. As this point corresponds to a point 21 days after the previous point used the wear and fouling the model is meant to notice may be the reason for the offset. To look into this even further a load period in which the boiler is run at high power output during a long time is analyzed. In figure 19 the output temperatures of super heater 1 are displayed

Figure 19: The steam temperature exiting super heater 1

Figure 19 shows how the temperatures of the steam differ from those calculated for in the model. During this period in February the boiler load was closer to 11 MW one MW above the design load. As mentioned in section 1.2.3 the model is expected to produce higher efficiencies than those that should appear when having run at high loads for longer periods. This as the model is designed for early fall after a long dormant period. This is clearer when looking at figure 20 displaying the temperatures exiting super heater 2.

(47)

-37-

Figure 20: The steam temperatures exiting super heater 2

(48)

-38- 7.2 Troubleshooting

As Hedemora Energi utilize oil fired boilers as backup for boiler 1 the costs of the two different fuels in use is highly relevant to evaluate the economic impact of downtimes of the same. In table 9 a simple cost evaluation of the downtime, starting at the 7th_{of December 2015 and ending at the 9}th of December, is displayed.

Cost of biofuel Cost of oil Downtime Intended Load Production cost

180 SEK/MWh4_{700 SEK/MWh}5 _{50 h} _10MW _{260 kSEK}

Table 10: A simplified cost analysis of downtime

As can be seen in table 9 the cost of switching to oil for about 2 days amounts to approximately 260 000 SEK. While stops such as this, by analysis of measured data, occur about 4-5 times a year the accumulated cost annually becomes about 1.3 MSEK. As the model developed in this thesis is intended for diagnostic purposes noticing some of these downtimes and possibly preventing them could be lucrative and environmentally friendly simultaneously. In figure 21 the load period preceding the downtime shown in table 9 is displayed.

Figure 21: The load period preceding and during the downtime in early December

As can be seen in figure 21 the load period preceding the downtime is subject to several dips wherein controllers may have tried to regulate certain error occurring in the boiler. At intervals of 8 hours the different hourly periods are run in order notice any discrepancies that, if increasing, may indicate faults requiring action that may render downtime unnecessary.

(49)

-39-

In figure 22 the steam temperatures exiting super heater 1 are displayed during the period preceding the downtime. This at as even intervals as possible to avoid plotting hours at which steam sweeps are performed.

Figure 22:The steam temperatures exiting superheater 1 at even intervals, 4 days preceding downtime

Noticeable here is how even though a large temperature difference between model values and measured values is present during the entire period the values diverge suddenly after 11:00, 5th_of December.

In order to further look into this specific case, the steam temperatures for the second superheater are also plotted, this in figure 23.

(50)

-40-

Figure 23: The steam temperatures exiting superheater 2 at even intervals, 4 days preceding downtime

As can be seen in figure 23 a similar trend as in figure 22 is present. This as the temperatures follow a similar curve for modelled and measured values until they diverge to a much higher degree at the 5th_{of December. As both graphs show convergence in retrospect the differences are shown to be} larger closing in on the downtime. The beginning of shutdown is displayed to show the amount of divergence at a maximum value. The shutdown begins at the 7th_{of December. This indicates that} some regulatory measure was taken as the largest divergence occurs. i.e. the bottom temperature of steam. To note here is that no steam sweep is performed during these hours. Thus, not explaining the drops in temperature.

One detail to note that during the entire period the pressure losses of steam calculated for in the model never differed more than a single percent compared to the measured values.

CHP process evaluation, optimization and troubleshooting in OpenModelica

CHP process evaluation,

optimization and troubleshooting

in OpenModelica

Abstract

SAMMANFATTNING

Contents

List of Figures

List of Tables

Nomenclature

ACKNOWLEDGMENTS

1 INTRODUCTION

2 MODELLING STRATEGY

3 The Concept of Design and off-design

4 Thermodynamics

5 Solving under design conditions

6 Solving under off-design conditions

7 Results