Blending into the mix
Successful coal blending requires finding a careful balance between fuel costs and plant performance. A recent study of a Malaysian power plant shows how utility TNB could reduce fuel costs while avoiding boiler operating problems normally associated with firing low-grade coals.
R. G. Adams, Dr. W. H. Gibb,
Dr. Kamsani Abdulah Majid,
Tenaga Nasional Research and Development,
In September 1998, Malaysian utility Tenaga Nasional Berhad (TNB) commissioned Power Technology to investigate coal blending at its Kapar power station in Selangor, Malaysia. After just six months of testing and simulation, Power Technology found that TNB could make savings on its fuel costs at Kapar while maintaining or even improving the plant`s operation.
The firing of coal blends is not new to Kapar. During the early 1990s the power station undertook combustion trials on units 3 and 4 with a blend of low quality coals. These were successful and led to more detailed combustion trials and boiler performance analysis in 1995. Following these trials, normal operation of the boiler continued with a variety of coals and low ratio coal blends 20:80 of two alternative coals (Merit Pilla or Tanito Harum). The blends have been used with varying success, but problems have been experienced, including boiler fouling, boiler steam pressure hunting and load limitations due to low mill outlet temperatures.
Some of these problems were thought to be the result of variations in the coal blend due to the current method of blending and known plant constraints, the most serious of which are insufficient primary air capacity and milling capacity constraints. The existing method of blending at Kapar uses simultaneous reclaim from separate coal stockpiles. This provides only a crude coal `mix` due to the discontinuity of operation with equipment such as bucket wheel reclaimers. This results in very conservative blends being used on the boilers which limit coal handling capability and does not exploit the full potential of cheaper alternative coals.
The need for improved coal blending at Kapar power station led engineers from Power Technology and Tenaga Nasional Research and Development (TNRD) to develop a blending strategy and identify future coal handling equipment changes. TNB wanted to widen the range of coals burned at Kapar to give the best overall benefit in terms of operating costs and increase the coal yard throughput to supply two new 500 MW units as well as maintaining the coal supply to two existing 300 MW units.
Kapar power station is a 2220 MW thermal power plant owned by TNB and operated by its subsidiary, TNB Generation. It comprises a mixture of coal, oil and gas-fired capacity including four 300 MW multi-fuelled units commissioned in the late 1980s which operate primarily on coal and two new 500 MW coal-fired units, the second of which is due to enter operation this year.
Power Technology`s task was to identify what coal blends (i.e. what percentage of low-grade coal) the boiler could tolerate and still give full load without violating plant constraints. The range of coals was to include all six current contract coals as well as future coals from the spot market.
The work comprised two main components. The first was a Coal Quality Impact Model (CQIM) analysis of the effect of different coals and coal blends on the combustion performance and economics. The second component was a performance analysis of the coal yard handling facility to determine whether the required blend accuracy can be consistently obtained and if the annual coal throughput can be achieved. The latter was done using Coal Handling Simulation (CHAS), a recently developed in-house software package.
The project team comprised two engineers from PowerGen`s Power Technology department and one from TNRD. Power Technology is the centre of engineering and scientific expertise within PowerGen, a UK-based power utility with a world wide interest in both generation and power distribution. TNRD is the research and development department within Malaysian utility TNB with offices near Kuala Lumpur. It undertakes a wide variety of work for TNB on a corporate basis.
A unique tool
There are a number of tools available for assessing the engineering impact of fuel quality on specific elements of power plant. However, CQIM is believed to be a unique tool since it determines the economic impact of variations in coal quality on the operation of an entire power plant.
The model considers all of the costs associated with burning a coal on a particular boiler under a selected operating regime. It considers delivered fuel price, maintenance costs, replacement generation costs associated with loss of capacity or availability, thermal efficiencies, auxiliary power requirements as well as ash disposal charges.
The model may be used to carry out long-term economic analyses over any desired period. The capital cost of any proposed plant modifications such as blending plant can be incorporated into the analysis. It can be used when very limited data on the specification and performance of a power plant is available, in which case default data built up from a large database of design and performance data is used. For Kapar, extensive data was available from previous tests to calibrate the model.
A full CQIM plant model contains an extremely detailed description of all the fuel-related elements of a power plant and can be used as a process model in its own right. The package also features a comprehensive maintenance database based on many years of historical data to allow the cost of variations in wear rate on plant to be determined. This database may be customized if more detailed local data is available.
The model is particularly good at evaluating the interaction of fuel quality effects, which are well understood, such as grindability and efficiency loss due to moisture. In other areas, where the fundamental processes are less well understood (e.g. slagging, NOx formation), the predictions from the model need to be reviewed critically and, if necessary, overwritten.
Power Technology has developed a bulk handling simulation facility designed to enable accurate modelling of a coal storage yard and conveyor systems. This was developed to assess the capability of coal handling plant designs to ensure that key items such as ship unloaders, conveyors and storage bunkers are correctly sized.
The user can define the bulk handling system interactively, the required operations and the shift operating pattern. A typical study covers an operational period of three months, simulated in one hour steps, to determine the berth occupancy rate, conveyor usage, boiler house bunker suitability and the adequacy of the shift operating pattern. The results are presented to the user numerically and graphically.
A typical CHAS screen is shown in Figure 5. The shipping, coal pile, boiler bunker and power plant details are shown on the left hand side and the simulation results plotted in the right hand window. At the bottom right hand side of the page is a summary of the simulation. This includes information about the shift pattern, period between ships arriving, the number of times operations are delayed due to conveyors being unavailable, conveyors utilization and berth occupancy.
To establish the most suitable blend ratios, the CQIM model for Kapar was run for a succession of blends from 100 per cent Merit Pilla to 100 per cent Blair Athol in ten per cent steps. The analysis was complicated by the fact that all coals have a range of properties and therefore a matrix of cases is used. Two problems emerged with high ratio Merit Pilla blends: a high mill throughput exceeding the design capacity, and boiler slagging.
Both of these can lead to de-rating of the boiler. With typical coal properties a 50:50 blend could be tolerated, but with worst case coal properties this might have to reduce to a 20:80 blend. Hence the target blend would need to be related to the as-load port or delivered coal properties.
A simple methodology was developed to identify the blend ratio with regard to mill constraints. This was based on three coal quality parameters that determine milling capacity, namely net heat content (net calorific value, NCV), moisture content, and grindability as defined by the Hardgrove Index (HGI).
A sensitivity analysis was performed by running CQIM for a variety of coals, both typical and worst case. Regression analyses were then performed to derive lines to define the CV and moisture content that equates to this mill loading at each of the HGI levels as shown.
For any potential coal supply to Kapar, a point can be plotted on Figure 3 based on the specified CV and total moisture content. Provided this point lies above the relevant HGI line then the required milling capacity for full load can be achieved. If the point lies below the line then mixing with a higher quality coal would be needed to ensure full load.
A distinction is made between the target blend and the achievable blend, to account for the blend method and any operational difficulties that are encountered with large scale bulk handling equipment. It is not uncommon for bucket wheel reclaimers to deliver a widely varying quantity of coal due to the discontinuous nature of its operation. This is not normally a problem with neat, single-supply coals. But blending by running two reclaimers simultaneously can produce a widely varying result and this process should correctly be termed `mixing`.
Hence a target blend of 50:50 could result in a wide normal distribution with a standard deviation from 40:60 to 60:40. Kapar has experienced this level of variation with previous blending trials and would normally result in the target blend being set down by 20 per cent to avoid pockets of blended coals which exceed 50 per cent of the alternative coal.
To maximise the benefits of a blend containing an alternative cheap coal, the narrower the blend variation the better. This allows the target blend to be set closer to the maximum identified by CQIM. A narrower blend variation can be achieved by adopting better blending methods than simply using two reclaimers.
The impact of the two blending methods is illustrated in Figure 4. The target blend is 50 per cent of the alternative coal. It can be seen that if crude blending is used the target must be offset by ten per cent to ensure that no part of this mix exceeds 50 per cent. However, if an improved blending method is used the target blend can be set closer to that desired, thereby reducing fuel costs without exceeding the 50 per cent limit.
Several coal blending methods were simulated to determine whether suitable methods could be found to produce an accurate blend while maintaining the required coal plant throughput.
Layering the stockpile is best suited to situations when the ratio of stockpile capacity to shipment size is large. This enables the shipment to be spread out thinly to give a good blend when it is reclaimed. Coal systems with a low ratio (i.e. less than 10:1) would find it more difficult to blend with this method because of the depth of each layer. Kapar is one such power station because of the relatively small coal stockpile area and the fact that it has a linear coal storage system that makes the layering option less attractive.
The option of controlled double reclaim was investigated. This is where two reclaimers operate in unison with some form of linked control. This is best suited to blend ratios which match the relative capacity of the reclaimers. All the reclaimers at Kapar are of a similar size hence a 50:50 blend would be best suited. Consideration was also given to using a flat back reclaimer, which would work in unison with a large bucket reclaimer and have its delivery rate controlled to match. This would suit a lower blend ratio due to the need to keep the flat back supplied with coal using mobile plant in an area constrained by the fact that it is a linear pile.
Kapar has a slot bunker used as a dry store. The formation of the blend in the dry store by double reclaim or controlled discharge from the slot bunker using paddle reclaimers was another method of blending considered and simulated.
A summary of the coal handling study results is given in Table 1, showing the blend accuracy and conveyor free time for various blending method and power plant load factors.
The table shows that for some cases the free time associated with the most critical conveyor in the system can reduce significantly by virtue of the need to blend. The amount of lost `conveyor free time` varies with blending method. The best method is one which gives accurate blending but will allow a high power plant load factor without reducing the conveyor free time to the point where routine maintenance cannot be undertaken. For Kapar this means that the last two options are favoured, i.e. storing the two coals in each half of the slot bunker and using the paddle feeders to form the blend or double reclaim using a flat back reclaimer coupled to a large bucket reclaimer.
The CQIM study showed that the proportion of cheaper coals could be increased from 20 per cent up to 50 per cent, provided each coal shipment was adequately sampled. The CHAS study demonstrated that the use of a flat back reclaimer or simple modifications to the dry coal store would allow accurate blending whilst allowing a high load factor to be maintained on the power plant.
Figure 2. The Kapar coal handling system
Figure 3. Prediction of milling plant capability
Figure 4. The impact of two different coal blending methods showing the amount of alternative coal in the blend