GC VAN TONDER, WN BOTHA,
Bara Industrial Plant Consultants, Engineering Division
9 Mount Road, Amanzimtoti, Durban, South Africa, 4125
Email: cornelius@baraconsultants.com
Contact Number: +27 82 557 3759
ABSTRACT
Package boilers are widely used for heating and process steam applications. Feedstock for these boilers has conventionally been A grade peas, which comes at a high cost. Lower grade coal and discard coal are available within the mining industry and at a lower price. Discard stockpiles pose an environmental hazard for mining companies and useable energy remains within these stockpiles. Investigation is needed to effectively combust lower grade and discard coal in fluidised bed boilers, while maintaining environmental restrictions. Evaluations were done to optimise package boilers and evaluating the use of lower grade feedstocks. A rand per tonne of steam produced value were used to quantify the effects within this study. Investigation was performed on a JTA MK2 boiler with an operating efficiency of below 60% and combusting A Grade peas. The cost of steam production was R311 per tonne of steam. It was found that fluidised bed boilers can effectively burn discard grade coal. Most effective size rating is grains (9mm to 16mm). This effectively lowered the rand per tonne of steam produced to R178.
Key words: Combustion optimisation, alternative feedstock, low HHV feedstock, Waste and effluent handling
INTRODUCTION
Fluidised-bed boiler has been in general use throughout the manufacturing industry for numerous years. Due to the ease of manufacturing and installation Fire Tube boilers are widely used throughout manufacturing industries to produce process steam and low yield electricity generation [1]. Fire Tube Boilers are of a fire-tube shell design with hot flue gasses being passed through tubes and heating water surrounding the tubes to saturated steam. These boilers burn a wide range of feedstock ranging from natural gas, oil and biomass [1]. This study specifically evaluated Fire Tube Boilers and fluidised-bed type package boilers. Studies have been done to look at the financial viability for large scale water tube fluidised bed boilers for power generation on discard coal [2]. This study however aims at smaller consumers in terms of package boilers and the viability thereof. According to a market survey more than 4000 Fire Tube package boilers exists within African. Department of Energy and Worldometers indicate that South Africa are ranked as the 7th highest consumer of coal in the world. South African coal usage accounts for approximately 17.8 % of consumption with an approximate annual production of coal 277 951 564 tons annually. This includes both bituminous, sub-bituminous (brown coal) and hard coal (Anthracite) [3, 4].
South-Africa coal reserves with the biggest consumers being of sub-bituminous coal (power generation and synthetic fuel), typically with a top side size of 100 % passing 25 mm and a calorific value below 24 MJ/kg. While the second largest consumer are anthracite coal, also consumed locally in a size range of 100 percent passing 25 mm and a calorific value of around 27.3 MJ/kg. Duff or fine coal are mainly consumed in calciner operation of cement factories and are typically discard from processing of anthracite coal for industrial applications. Currently a large international demand also exists for a similar size and high-quality grading coal as consumed by industrial users, making it financially a big expenditure for industries locally as the market demand is high. A Grade pea size coal prize showed an increase of 200 % in the last two decades. According to North et al, 2015, this large export market of anthracite type coal, requires beneficiation processing on mines to reduce ash content of coal and subsequently increase the coal calorific value. Beneficiation produces discard coal, and coal slurries that are stockpiled on mines [2].
According to the South African Department of Energy, approximately 60 million tons of Discard coal are produced annually and has accumulated to approximately 1 billion tons over the South African Mining life [3]. This leads to large amounts of energy that is unutilised, management of these discard heaps also leads to environmental concerns. These heaps pose the risk of spontaneous combustion and has become a source of acid rock drainage contaminating underground water supply [2].
The beneficiation waste streams that are stockpiled are classified as, discards, duff and slurries. Discards are specifically of a higher ash fraction coal, and have a lower calorific value range [2].
Considering the current worldwide trends in seeking cleaner alternatives and economic downturn due to the pandemic, the way local industries utilise energy needs to be reconsidered. For smaller manufacturing industry however retrofitting and clean coal technologies are not financially viable options to attain targets, nor thus the current economic environment makes it quickly attainable. It is under these conditions that this study was undertaken, knowing that large amount of environmentally hazardous discard coal exists, the fact that fluidised bed boilers should be able to handle these quality coals operationally better and the economic climate pre and post COVID that necessitates industrial savings.
During the 1980’s similar type investigations were conducted by the CSIR but were abandoned later on.
The boiler under consideration in this study was a John Thompson Mk2 coal fired fire-tube spreader-stoker boiler with a maximum steam production rate of 6.3 t/hr. The plant is in Hammersdale, Kwazulu-Natal, and is about 500 km from the Highveld area.
METHODS
The methodology in this study consisted of first determining the feedstock characteristics, finding sourced coal that met the discard and financial requirements and classifying that characteristics. Secondly, theoretical modelling took place to determine whether the JTA MK2 boiler could achieve the desired steam production with the quality of coal (Stoker rating basis). Baseline boiler performance modelling were then performed, after which criteria for successful testing of discard coal were developed, combustion testing and optimisation studies were then performed. The following Sections outlines the main principles used under each of these methods.
Feedstock characteristics
The feedstocks used in this study were between A grade peas and discard coal of which the qualities are shown in Table 1. All feedstock was characterised using ISO sampling and SANAS accredited testing methods.
Table 1 Feedstock Characteristics
|
A Grade Peas (As received basis) |
Discard Coal (As received basis) |
||
|
Calorific Value [MJ/kg] |
27.3 |
Calorific Value [MJ/kg] |
21.3 |
|
Total Moisture [%] |
3 |
Total Moisture [%] |
3 |
|
Volatile Matter [%] |
25 |
Volatile Matter [%] |
22 |
|
Ash Content [%] |
12 |
Ash Content [%] |
23 |
|
Sulphur [%] |
0.5 |
Sulphur [%] |
0.9 |
|
Fixed Carbon [%] |
577 |
Fixed Carbon [%] |
54 |
Note that the coal significant variances are picked up in terms of Ash content, Calorific Value and volatile content. Volatile content is directly proportional to burnout time of coal and is one parameter that was considered within this study. The feedstock characteristics shown in Table 1 were used to further develop theoretical models and do the necessary stoichiometric calculations for ideal combustion which were used for combustion optimisation purposes.
Base line performance – Theoretical Model
To effectively quantify boiler operation baseline performance modelling were performed. These investigations were based on the first principle thermodynamic modelling that were further solved using the Newton-Rapson method to determine the constraints as shown in Table 2 for various coal qualities.
Table 2 Baseline Thermodynamic Model Output
|
Output [A Grade Peas] |
Output [Discard Coal] |
||
|
Coal Density [kg/m3] |
830 |
Coal Density [kg/m3] |
830 |
|
GHV [MJ/kg] |
27,3 |
GHV [MJ/kg] |
21,3 |
|
Grate area [m2] |
2,4 |
Grate area [m2] |
2,4 |
|
Grid Speed [RPM] |
5,2 |
Grid Speed [RPM] |
5,2 |
|
Grid speed [m/h] |
3,15 |
Grid speed [m/h] |
3,15 |
|
Bed Thickness [m] |
0,13 |
Bed Thickness [m] |
0,18 |
|
Grate width [m] |
1,21 |
Grate width [m] |
1,21 |
|
Guillotine Area [m2] |
0,16 |
Guillotine Area [m2] |
0,2178 |
|
Coal Flow [kg/h] |
411,62 |
Coal Flow [kg/h] |
569,93 |
|
Rating [MJ/m2h] x103 |
4,68 |
Rating [MJ/m2h] x103 |
5,05 |
|
Desired Rating [MJ/m2h] x103 |
5 |
Desired Rating [MJ/m2h] x103 |
5 |
Table 2 shows the model output for both the A grade pea coal that were being consumed by the plant, with a 27.3 MJ/kg calorific value and discard coal with a 21.3 MJ/kg calorific value. From the Table it can be noted that a desired rating of 5 [MJ/m2h] x103 is required for Boiler MCR conditions, and that it is theoretically achievable under both conditions, with variable bed thicknesses.
Initially discrepancies between this model and the boiler were noted. As the actual plant control system were incorrectly set up.
Figure 1 shows the theoretical model output for speed, bed thickness at various loading conditions, this Figure gives valuable insight into theoretical combustion control requirements on a fluidised bed boiler.
Figure 1 Model output steam production versus bed drive speed
Figure 1 gives a calibratable output that were tested at the given feedstock to calibrate model outputs to reflect the plant. This first principal methodology enables end users to make deep engineering prediction in system responses and can further be utilised to diagnose system faults and discrepancies once the model is calibrated to reflect the current operation.
Base line performance – Actual boiler operating conditions
The base line plant performance was determined by taking plant data over time and calculating a steam produced to coal used ratio. This ratio gives an indication of the actual boiler efficiency over time. The importance of establish this value is so that there would be tangible criteria to measure any changes against.
The method to establish the baseline, was to measure steam flow with a Spirax Steam flow module that measures hourly production rates, coal flow was determined with from the effective area, bulk density and velocity of the bed. These values were then used to determine a steam produced versus coal usage ratio, efficiency ratio. This provided the base line performance then in which further improvements could be established. Theoretically with lower coal quality efficiency will drop.
Combustion optimisation
Combustion optimisation is essential in terms of ensuring the best results for any given combustion scenario. This was done by firstly evaluating the control set up, correcting all control parameters to reflect the plant conditions, and secondly the combustion controller were placed on manual stack readings were taken, air flow set to the stoichiometric amount plus excess air as determined by the combustion models and feedstock characteristics. It was critical to ensure that prior to changing the feedstock, that the boiler draught group and control be fully understood and optimised.
Sizing evaluation
Coal Sizing plays an important role in fluidised bed combustion, as sizing relates to burnout time of coal. Conventionally size grading for these type boilers have been earmarked as peas (6mm to 25 mm) sizing, with most of the coal sizing distribution laying between 10 mm to 20 mm. The problem with pea size coal or larger than pea size coal would be that when the volatiles drop, ignition conditions cannot be maintain within the flame arch of the boiler. This could lead to combustion losses, the lower the volatile content of the coal smaller sizing would theoretically aid in reducing the burnout time. However, a size distribution with too much fines would clog up the bed, be rejected into the boiler riddling hoppers, and causes blowholes during combustion. Within this study various sizes were evaluated including small nuts (larger than 25 mm), pea size coal, low quality coal with 100% passing a 23 mm sieve, and A grade grains (6 mm to 16mm) with the majority of the sizing distribution being between 10 mm and 12 mm.
Criteria for lower quality feedstock
The first step in burning a lower quality coal was like indicated above. To determine thermodynamically, the heat required by the boiler at maximum productions rate. From the heat required, it can then be determined if the discard as shown in Table 1 would be a suitable grade. Which would satisfy the heat and energy requirements of the boiler while reducing the operating costs. The information from the review would determine that coal burn rate, required air flow and bed feed-rate. In conjunction with this analysis a combustion for that specific grade feedstock would be needed to be done and control settings had to be reviewed to facilitate the lower grade coal. If a low-quality coal is inserted into the boiler and insufficient combustion takes place due to incorrect control set up it could result in the study drawing an incorrect conclusion.
Criteria for successful results on discard coal were developed and are listed below:
- Lower grade feedstock should be able to sustain production requirement.
- Should stay within environmental constraints.
- Should maintain ignition.
- Should not overburden the mechanical plant or the operating personnel.
Results and Discussion
Baseline performance – Theoretical and Actual
For this specific boiler, the water was heated from 90˚C up to a saturation conditions of 180˚C at 1000 kPa and this was used to determine the amount of energy required from the coal. The require heat/energy was determined as 5 MW. The anthracite coal with GCV of 27.3 MJ/kg was used to calibrate the theoretical model, baseline performance was then established on actual plant data as mentioned in Section 2 and theoretical data. Actual plant data are shown in Figure 2 on a daily average.
Figure 2 Actual plant performance baseline
From Figure 2 it can be noted that on average within the first 29 data points, a high efficiency reading is observed (average of 8 TS:TC which equates to an 80% and above boiler efficiency). When correlated with the theoretical models it was noted that the theoretic model shows an efficiency of 40% based on the coal usage and a TS:TC ratio on average of 4.5.
Further investigation was done in the plant coal accounting practices and discrepancies noted. The control system set-up was then investigated. And noted input of the control system showed the grate width as 0.71 m as opposed to an actual plant value of 1.2m. This explained the coal accounting discrepancies. The value was change to the measured value and other control parameters changed to reflect the current plant set up as well. From Figure 2 the drop in baseline performance values can then be noted from data point 29 and onwards. The theoretical model and actual plant conditions now correlated with each other. A production of steam costing value of R310 per ton of steam produced were then determined and used as the baseline value for the result of the study. In essence this meant that the plant was running inefficiently even with the high-quality coal. This inefficiency was attributed towards, carbon in refuse loss (Operating principles) and dry flue gas loss with stack temperatures exceeding 200°C for over 90% of the operational time of the boiler.
Combustion Optimisation
Figure 3 shows the combustion control performance during baseline investigations.
Figure 3 Combustion control response
From Figure 3 it can be noted that the combustion control responses were erratic. Both stoker speed and draught group had cyclic loading characteristics under normal operation, considering that this plant have a steam accumulator upstream, the result were unexpected. Upon this finding PID control tuning were performed, coal bed characteristics changed within the control system and the response factor change by which the Forced draught fan supplies air in increase of the stoker speed. Note that the control principle here is Induced draught leading.
Figure 4 shows gas analysis versus specific plant changes.
Figure 4 Combustion Optimisation
The combustion optimisation started by first determining at around what oxygen (O2) and carbon monoxide (CO) percentage incomplete combustion started to take place inside of the boiler. Low O2 and high CO in the flue gas stream is an indication of incomplete combustion. This was then correlated with the boiler input (bed height, Air: Fuel ratio, excess air) and output (steam flow, fan speed, stack temperature, ash produced). This can be seen Figure 4.
Initially the boiler was starved from air by setting the air: fuel ratio to -20% and slightly reducing the excess air from 30% to 10%. The bed thickness was set at 120mm. During this period, it can be seen from Figure 4 that the O2 concentration drop significantly until it reached a point where it measured around 1-2%. It can also be seen at this point that the CO reading were at it is highest and was measured at around 25000 ppm. Furthermore, the colour of the flue gas turned dark black and high carbon-in-ash was noted (See Figure 5). What was also noted was that the system had a delayed response of around 15 minutes meaning that changes in control settings that were made to the amount of air introduced to the system would only be seen 15 minutes afterwards. This is mainly due to the residence time in a fluidised bed boiler.
Figure 5 Ash sample during combustion optimisation
From Figure 5 it can be noted that large amounts of unburnt coal are present in the ash. The next step in the test was to find the optimal control settings for the current coal. According to the theoretical model it was seen that in theory the demand of the boiler can be met with bed thickness of below 100mm. The bed thickness was set to 96mm and oxygen trimming was done using the flue gas analyser and adjusting the air: fuel ratio and excess air settings. After numerous adjustments, an optimal O2 percentage was found of around 7-8 %. At this value complete combustion of the coal took place and the boiler demand was met. Note in Figure 6 below the presence of unburnt coal decreased significantly.
Figure 6 Ash sample after combustion optimisation
From Figure 6 It can be noted with the boiler operating close to ideal conditions the presence of unburnt carbon is significantly lower. This was also reflected in the TS:TC ratio being redetermined. This value increased from on average 4.5 during baseline evaluations to 7 on average during higher production loads and 6.3 on average during lower production conditions. The snapshot shown in Figure 4 shows a reduction in rand per steam produced from R311 to R218 based solely on achieving the ideal combustion conditions. However, this was found to not be long term sustainable as operating culture would change settings again in the control system, which means this exercise needed to be repeated.
Sizing evaluation
Various coal sizes were evaluated in this study, including grain, pea size and small nut coal. A low-grade small nut type coal were evaluated first and it was noted that combustion conditions could not be maintained, ignition would be achieved after which the flame would not be sustained within the arch. Subsequently leading to a decrease in the flame arch temperature until incoming coal thus not ignited.
The evaluation into combustion of small nut size discard coal looked at changing multiple variables to maintain the flame within the arch. This included slowing the stoker speed, changing the velocity within the furnace until it did not correlate with the theoretical model. I.e., under the condition’s combustion were maintained it would not be able to maintain production requirements.
Discard of a size ranging of 100 % passing 23 mm were then evaluated, this showed that successful combustion could be achieved and maintained and that it could sustain the production requirements. A similar exercise were done on grain size coal to determine in both low grade and high-grade coal whether grain sizing would improve burnout efficiency as the burnout time should significantly reduce with the smaller surface area.
Grain size coal showed the highest efficiency of all the coal qualities, both in terms of the low-grade coal and the higher-grade coal qualities. With efficiency improvement moving from 45% to 84% in both cases. With the discard coal sufficient output were achieved to maintain the production requirements. Sootblowing and de-ashing did increase as would be expected. It was still well within manageable levels.
The boiler control setting was changed according to the results obtained by the test to ensure complete combustion of the low-quality coal. A load of around 40 tons of the test coal was delivered to the plant to conduct the test.
Further testing was also conducting in blending of different coal streams, A grade coal with discard coal, it was noted that under all blending conditions with sufficient combustion optimisation that blended options were also viable to maintain production throughput.
Conclusion on Discard coal
From the test conducted it was proven that with the correct size grading and combustion control optimisation discard coal can sufficiently maintain production and burn efficiently enough to be a financially viable solution. Figure 7 shows the steam produced costing.
Figure 7 Steam produced costing
From Figure 7 it can be noted that a significant decrease in plant expenditure were achieved when moving to discard coal. This equates to R4.1million rand annual saving for this specific plant assuming full load production throughout the year.
DISCUSSION
The results obtained during the tests were very insightful in how optimization of fluidised bed packaged boiler can be done. One of the major factors that were picked up prior to testing of the coal is to analyse and review the boiler control system. It is critical to first understand how the control system of the specific boiler operates and more specifically how it controls the amount of fuel and air being supplied to the boiler. Any errors or discrepancies will cause incorrect accounting (as seen) and ineffective combustion. These issues must be sorted out first prior to changing the feedstock of the boiler. The results also showed that a cheaper lower-quality coal can indeed be burned on these boilers but it essential that the coal adhered to the correct sizing.
Research show that a lot of mines in South Africa have large stockpiles of high/low-quality discarded coal due to coal beneficiation process. The general perception has been that packaged boilers must operate on A-grade peas size coal and that it has created a competitive market for it at high prices. Using combustion principles and plant knowledge these types of high/low-quality discard coal can be burned on boilers and save plants significant amounts of money in tough financial times. This, in conjunction with the large amounts of discard coal found in South-Africa, could revolutionise the packaged boiler industry as it would create a new coal market while reducing boiler operators input costs.
References
[1] Singh, A., Sharma, V., Mittal, S., Gopesh, P., Mudgal, D. and Gupta, P.: "An overview of problems and solutions for components subjected to fireside of boilers" International Journal of Industrial Chemistry Vol. 9, No. 15, March, 2018
[2] North, B., Engelbrecht, A. and Oboirien, B.: "Feasibility study of electricity generation from discard coal" Journal of the Southern African Institute of Mining and Metallurgy, Vol.115 No. 7, July 2015, pp. 573-580.
[3] Department of Energy, http://www.energy.gov.za/files/esources/coal/coal_discards.html, Date of Access: 2 October 2020
[4] Worldometers, https://www.worldometers.info/coal/coal-consumption-by-country/, Date of Access: 2 October 2020