World’s First Coal to Biomass Conversion Using Advanced Wood Pellets

In 2015, a coal-fired unit in Canada was successfully converted to burn thermally upgraded wood pellets. The first-of-a-kind project was executed with a capital cost of about $30 per kilowatt and has demonstrated its feasibility at full scale.

Overhead View of Advanced Wood Pellet Loading (September 2013)

Ontario Power Generation (OPG) has been actively evaluating the use of biomass to decarbonize the production from its coal fired assets for some time. In recent years, this focus has shifted to the testing of second generation wood pellets to prove the ability to employ these new fuels safely and effectively within an existing coal fired unit. OPG has recently executed a project to convert Thunder Bay Unit 3 from coal to 100 percent biomass firing using steam exploded wood pellets.

In February 2015, this unit entered service on the new fuel, making it the first such unit worldwide to employ thermally upgraded wood pellets. The unique properties of the steam exploded pellets enable the use of outdoor storage and handling, similar to the baseline coal firing case. The ability to store, handle, pulverize and combust the pellets with only minor modifications to the plant systems and procedures has enabled the use of a low capital cost project approach that is not possible with conversions that employ traditional white wood pellets.

The conversion of Thunder Bay Unit 3 was executed with a capital cost of approximately $30/kW and has demonstrated the feasibility of this approach at full scale. Observations from the fuel evaluations, test burns and commissioning of this first-of-a-kind project are discussed.

OPG began a significant biomass test program in 2006, in direct response to Ontario provincial regulation O. Reg. 496/07 (August 2007) — “Ontario’s Cessation of Coal Use”. This regulation effectively prohibits the use of coal as a fuel to generate electricity in the province of Ontario as of December 31, 2014. Engineering studies were conducted for all of the OPG coal-fired stations to evaluate the feasibility of converting to natural gas, wood pellets or a combination of these two fuels. However, the economic downturn of 2009 reduced electricity demand in the province. This fact, combined with the addition of several gigawatts of NGCC capacity in the province resulted in a very limited role for the large coal-fired station at Nanticoke (4000 MW) and Lambton (2000 MW).

Conversion efforts then focused on the station in northwest Ontario at Atikokan. This unit is modestly sized at 205 MWe and is in an area of vast forestry resources, making it an excellent candidate for conversion as a peaking unit. The Atikokan unit has been completely converted from coal to biomass firing and entered service in July 2014. The total cost of the Atikokan conversion was about $170M CAD, yielding a specific capital investment cost of some $800/ kW ($CAD).

Concurrent to the efforts at Atikokan, OPG has been studying the potential of upgraded second generation wood pellets. The generic term “advanced wood pellets” describes an array of pelletized woody biomass that have undergone some type of thermal upgrading. The pre-treatment processes and specifics vary but include the use of torrefaction, hydrothermal carbonisation and steam explosion. In all cases, the intention with advanced wood pellets is to provide a fuel with properties that are like coal.

White wood pellets are the dominant fuel choice for the coal-to-biomass conversions. Wood is a relatively clean fuel and the production and logistics pathways to deliver utility scale volumes were already under development to support other markets. However, the use of traditional white wood pellets in a conversion project does require a significant capital investment. Mandatory new systems include covered storage to protect the pellets from exposure to the elements and dedicated receiving and handling systems to control dust generation and mitigate the risk of fires and explosions. These projects also typically require either new milling technology or the modification of existing coal pulverizers to properly and safely handle the wood pellet fuel.

The aim with the development of advanced wood pellet technologies is to produce a biomass fuel pellet that can be employed using existing power station assets with only minor modifications. In this respect, the pellets should be stored outdoors without covered storage so that they can be received, stored and reclaimed from the yard using equipment and procedures developed for coal. This initial item can have an enormous positive impact on the capital cost of a cofiring or conversion project. However, the advanced pellet must be able to weather the elements so that excessive dust is not generated from degraded pellets when they are handled.

In December 2012, OPG was asked by the Ontario Ministry of Energy to explore the potential to convert Thunder Bay Unit 3 from coal to 100% biomass firing using a low capital cost approach. The Ministry had been previously informed of new developments with upgraded biomass fuels that might be utilised to avoid the high capital costs that are associated with a typical white wood pellet conversion pathway.

Thunder Bay Generating Station is in northwest Ontario, Canada on the north shore of Lake Superior. TBGS Unit 3 has a nameplate capacity of 163 MWe (gross) and includes a four-corner tangentially fired boiler. The boiler is equipped with five RP 783 pulverizers. The boiler was originally designed for Western Canadian lignite coal (Luscar) but was converted to fire Northern Powder River Basin (NPRB) coal in 1996.

As of 2013, OPG had already been evaluating upgraded biomass pellets for a period of about four years. Dozens of different samples and suppliers were analyzed and assessed through an internal OPG program and via collaborative efforts with other utilities. The areas of evaluation can be summarised into several categories: Weatherability, Dust Generation, and Milling.

The use of advanced biomass fuels will have other impacts on the unit, similar to the firing of any biomass based fuel. However, these areas highlight the unique differences between standard and advanced pellets. The observations by the OPG project team in each case are detailed below.

Pellet durability — also referred to as the pellet durability index (PDI) — is the standard indicator of the relative mechanical strength of a pellet. Durability can be determined via many techniques and methods. For the purpose of this discussion, we refer to the standard tumbler method as detailed in ISO 17831-1. The durability metric does not have any direct correlation to performance at full scale in an industrial setting. It is merely an objective, repeatable means to compare the relative quality of biomass pellets with respect to their mechanical strength and ability to resist degradation when handled.

OPG recognised that the standard durability test, conducted on air-dried, pristine pellets, was not a representative measure for pellets that would need to be stored outdoors without the benefit of covered storage. Earlier work by the project team had confirmed that exposure to water was the key mechanism leading to pellet degradation and the production of dust. The biomass power industry has used a number of different methods to simulate exposure to the elements, including climate chambers with high humidity environments and test rigs that simulate actual rainfall. OPG has adopted the use of immersion in water as a simple and objective test that can be conducted by any laboratory, utility or fuel supplier.

The simple weatherability test developed by OPG can be summarised as follows:

This method will yield a direct comparison of pristine pellet fines and durability with that following a simulated exposure to the elements. It also return a value for water uptake — the absolute increase in moisture content following soaking — that will be an important consideration when milling is discussed.

The obvious question then is how long should the pellets be immersed in water. OPG has conducted trials using durations from one hour to one week in length, tracking an increase in water uptake and pellet degradation with increasing time submerged in water. A duration of 48 hours has been selected by OPG as the rate of water uptake is observed to flatten after two full days of soaking. This level of exposure also provides significant differentiation between pellets of varying qualities in this important area of performance. OPG has also observed a good degree of correlation with lab scale soaking for 48 hours and actual results with full scale outdoor storage over a period of months. This latter result will be discussed further in the operational commissioning section.

Current lab scale methods cannot be expected to predict specific performance in the field, especially for a range of environmental conditions and storage periods. However, the application of objective lab scale methods such as durability – and especially weathered durability — are very useful to compare the relative characteristics of fuels in this area. OPG had already evaluated many thermally upgraded pellets in the 2010-2013 period and initiated a new program to test all promising advanced wood pellet fuels specifically for the Thunder Bay project. Weathered durability was a key metric in this testing, used to identify fuels that would tend to maintain their pellet integrity during outdoor storage in the challenging conditions of northwest Ontario.

Ten samples were tested in this program, eight of which were torrefied wood pellets and two produced via steam explosion. Given the first-of-a-kind nature of the project and the important link between pellet integrity and fuel handling safety, the project team sought to identify pellets with outstanding performance in this area.

This work confirmed the trend for steam exploded pellets to perform significantly better than torrefied fuels when exposed to water. The lab scale weathering of a typical torrefied pellet is compared visually with that of the steam treated pellets selected for the TBGS conversion in Figure 1.

Immersion of the torrefied pellets in water has had a clear and significant negative impact on the integrity of the pellets. Note that the fines metric is an indication of the mass of the sample that is already dust. Significant exposure to water can be seen to produce dust from the original pellets even before they are tumbled again to simulate handling.

With respect to the water uptake value, we suspect that the individual torrefied wood particles are quite hydrophobic. The high level of moisture absorption in the soaked pellet sample is likely due to water accumulating interstitially in the voids between particles that have separated as exposure to water decreases the strength of their bonds.

Some torrefied pellets do perform better than others in this regard and there has certainly been progress in this area in the 2013-2017 period. However, during the time when the fuel selection was being made for the Thunder Bay project, steam treated pellets were seen to have a clear advantage.

Wood pellets — including advanced wood pellets — are actually more problematic to handle when they are very dry as the pellets are more brittle and prone to produce dust when handled. In this respect, it is important to evaluate the amount of dust that is generated as well as the particle size of that dust. Finally, the use of water and surfactants as dust mitigation measures should be assessed. This scope of testing was executed on the steam treated pellets that were selected for the project based on their excellent performance in the weathering tests.

An external laboratory was engaged to conduct a third party evaluation on the advanced wood pellets as well as a typical PRB coal as a baseline. The method of dust generation used was a rubber-lined tumbler, operated for 48 hours. This testing also included the determination of the particle size ranges of any dust produced. Table 2 summarises these results.

These results confirm the favourable performance values cited earlier. The relatively small volume of very fine dust is particularly important as this can significantly reduce the risks associated with airborne dust generated during handling. The results for the AWP also compare well with those of the PRB coal. This baseline was important to the project team as the performance of the handling system on coal was both well controlled and well understood.

An additional set of tests was also conducted to evaluate the effectiveness of water and dust suppression agents to mitigate dust generation during handling operations. The lab scale Walker Wetting Test was employed to assess performance in this area. The results are shown in Table 3.

The BT-220W surfactant is the same wetting agent used at Thunder Bay GS for operation on coal. The Walker Wetting test results indicate a significant improvement should be realised when using surfactants to mitigate dust with the advanced wood pellet fuel.

The final major benefit of advanced wood pellets is their improved grindability, relative to standard white wood pellets. Several conversion projects (including OPG Atikokan GS) have successfully retrofitted coal pulverizers to handle white wood pellets. However, there are operational challenges, most notably with respect to milling capacity and the final delivered particle size of the wood dust to the burners can create issues downstream as well.

There is the potential to employ advanced wood pellets in an existing system with only minor changes to the physical mill and the operation of the pulverizers. Furthermore, there is an expectation that mill capacity (energy output) of the mill should be improved over the white pellet case and that the particle size of the pulverized wood dust should be significantly smaller, yielding a number of downstream benefits with pneumatic transport and combustion.

Even as early as 2013, the biomass power industry had realised that the standard Hardgrove Grindability Index (developed for coal) was not an appropriate metric to determine the relative grindability of biomass fuels, including advanced wood pellets. As of this writing, new methods are in development but no such tools were in place when OPG was evaluating fuels for Thunder Bay GS. It was decided to use pilot scale mill testing to assess the steam treated pellets in this area as OPG had already employed this technique for a previous white pellet evaluation at Thunder Bay.

Several campaigns of pilot mill testing were conducted at the Alstom Pulverizer R&D centre in Naperville, Illinois. This facility is equipped with a vertical spindle VR 31 mill that uses similar grinding technology to that employed in the full scale RP 783 pulverizers installed at Thunder Bay GS.

Parametric testing included the variation of the bowl rotational speed, roll-table clearance, roll loading pressure, fuel flow, air flow and classifier vane opening. The facility also offers the opportunity to physically modify the mill internals. In this case, the classifier static drum openings were increased to better reflect the full scale case at TBGS. Most importantly, tests were also conducted after the removal of the outlet venturi or discharge skirt (see photo on page 30). This modification resulted in the best performance of the pilot scale program and informed the project team of the potential to modify the Thunder Bay pulverizers in a similar manner.

The use of a pilot milling facility to test wood pellets had many advantages, especially regarding the flexibility to modify the physical configuration and the operational parameters. However, the nature of advanced pellets does result in certain observations that require some degree of interpretation. In particular, specific mill power was observed to be rather high, with values of about 20-25 kW/Mg. This is approximately twice the value that would be expected for coal grinding, using either a pilot scale or full scale mill. Fortunately, this result was actually much more favourable at full scale when grinding steam treated pellets in the modified mills at Thunder Bay.

A standard suite of fire and explosion risk testing was conducted on the steam treated fuel selected for the Thunder Bay conversion project. These results confirmed the increased ignition risk with biomass fuels, compared to the PRB coal baseline experience (Table 4). Concurrent with this effort, third party evaluations of the entire (coal) handling system were conducted to identify issues of concern and areas for improvement.

The review of the handling systems identified many maintenance issues in the areas of containment and dust collection. Furthermore, several components of the system were also highlighted that would benefit from modifications to improve safety by avoiding and mitigating the build up of electrostatic charge. These items will be discussed in the conversion project scope.

Pilot Scale Mill Classifier with Discharge Skirt Removed

The project team was eventually satisfied that dust and ignition hazards had been properly addressed by the proposed modifications throughout the system — with the notable exception of the final drop into the existing coal bunkers. Under certain conditions, the free fall of pellets into the metal bunkers could result in an electrostatic discharge with the potential to ignite wood dust in that confined volume. A dedicated study of this issue was executed to identify possible solutions to the problem.

The electrostatic discharge study determined that the risk of bulking discharges were sensitive to the relative humidity in the environment. The steam treated wood dust was tested in accordance with ASTM D257 — “Standard Test Method for DC Resistance or Conductance of Insulating Materials (Modified)”. This work was conducted at two different levels of relative humidity. The results are summarized in Table 5.

The study concluded that the theoretical threshold value for volume resistivity is 1010 Ω-m, indicating that bulking (or cone) discharges do not occur below this value. The project team elected to utilize the existing bunker inerting steam system to humidify the bunker volume prior to loading pellets. Humidity meters were installed in the target bunkers and fueling operations commenced after the bunker relative humidity was increased to 55+ %RH. This is likely the first such direct manipulation of humidity to control an ignition risk in the industry.

Safety has always been the number one priority at OPG and the Thunder Bay project maintained that focus. Given the first-of-a-kind nature of the conversion project and the desire to largely use existing systems to handle and fire the new fuel, the selection of an advanced pellet fuel with excellent performance characteristics quickly became the first critical decision.

The internal OPG evaluation process determined that the steam treated pellets produced by Arbaflame AS (Norway) yielded clearly superior performance in the important areas of durability, fine dust generation and weathering. The Arbaflame pellets were also observed to have clearly superior performance with respect to water uptake when exposed to the elements. Additional testing in the areas of fire and explosion risks, liquid-based dust suppression and pilot scale milling all gave expected or acceptable results.

Concurrent to this effort, OPG also collaborated with several European utilities with their own advanced biomass programs. Most notable among these was Vattenfall who had previously formed their Black Pellet Evaluation Program to investigate the use of advanced biomass fuels in their fossil fleet. Vattenfall had independently selected Arbaflame as the leading candidate and had already conducted several major field tests, including a large scale co-firing trial at their Reuter West station in Berlin.

Consideration of the OPG test results, combined with similar favourable experience from other utilities, resulted in the project team decision to procure 1000 metric tonnes of Arbaflame pellets for a test burn at Thunder Bay in the fall of 2013.

A significant amount of maintenance was performed on the existing fuel handling system, addressing items that were identified by third party walk downs of the equipment. In addition, several physical modifications were also executed, again in direct response to issues identified by pre-test safety studies.

The final significant equipment modification was with the pulverizers selected to handle advanced wood pellets during the test burn.

Two of the mills (3B and 3C) were modified to emulate the successful configuration adopted during the pilot scale mill testing. The discharge skirt (or outlet venturi) was removed from these mills to allow for a more expeditious path for fuel to exit the mill. The photo on page 23 shows the discharge skirt in the Thunder Bay mills prior to removal.

The roll-table clearances on these mills were also tightened, based on results from the pilot scale mill program.

No modifications were made to the pulverizer throats (vane wheel) prior to the initial field tests. This oversight will be discussed in the commissioning sections.

Reclaim Hopper Slide Gate Extension

Dedicated Grounding of Dust Collector Cages

The chemical analysis of the Arbaflame advanced wood pellets is compared with the baseline Northern Powder River Basin (NPRB) coal in Table 6. The chemical analyses of the two fuels are actually quite similar, with notable differences in the acid gas precursors (nitrogen and sulphur), as well as the ash and moistures contents.

Of course, the major expected differences in fuel performance are not covered in a basic chemical analysis. The handling and milling properties of the advanced wood pellets were the main focus of the test burn. The 1000 Mg test fuel volume was delivered to site in August 2013 and piled in the yard as shown in the photo on page ??.

This fuel pile was sampled over the duration of the field test period to determine the fines and pellet durability. This is summarized along with lab scale weathering results in Table 7. In the August-September 2013 period, the fuel was stored outdoors as pictured, without the benefit of cover.

As noted previously, several modifications were made to the existing handling system. In addition, the indicated classifier modification was executed on the target test mills only — mills 3B and 3C. These pulverizer levels are located near the bottom of the firing system (A-lowest through E-highest).

The first week of testing involved the testing of each pulverizer on advanced pellets, with supporting mills firing coal. The second week of testing employed both mills on pellets with the goal of evaluating pure biomass firing operation, including unit start up and shutdown.

The initial loading point for the advanced wood pellets was reclaim hopper #2. The handling operations in the yard were accomplished using normal mobile equipment (as for coal). The photo on page 16 shows a typical loading operation during the first week of testing. The dust cloud formed by the inherent fines in the fuel volume is very apparent. However, it should be noted that this was the only location where airborne dust was observed during the test program.

Relative Humidity Meter

Previous testing by Vattenfall has confirmed the ability to control dust formation at the initial loading point by means of a simple water jet spraying across the mouth of the loading hopper. This additional mitigation was not deemed necessary by the commissioning team at Thunder Bay as the level of dust and the extent of propagation was deemed acceptable.

The favourable dust and handling performance of the pellets was confirmed by sampling at all of the downstream transfer points. To facilitate this sampling, the entire conveyor system was tripped during the initial test day and cross belt samples were.

As noted here, the durability of the pellets is essentially constant throughout the handling system and appears to be unaffected by the falls through the transfer points. The fines performance is also good, indicating that dust generation is limited as well.

As discussed earlier, each of these transfer points was equipped with dust suppression sprays, employing either surfactants or water. The effectiveness of these sprays on the advanced wood pellets was excellent and did not result in a significant moisture pick up. Those results along with the fines data are summarized in Figure 2 on page 24.

In addition to fuel sampling, the test program included dedicated airborne dust monitoring in the fuel yard, all conveyor galleries as well as wearable personal breathing zone monitors for staff working in handling operations. This effort was conducted for pellet handling as well as for similar volumes of NPRB coal.

The air borne dust monitoring results were very encouraging, superior to those for the baseline case handling coal.

The personal breathing zone results were all well below the Ontario Occupational Exposure Limits (OEL) for softwood dust (5 mg/m3), averaging around 0.30 mg/m3 for workers in the fuel yard, including mobile equipment operators.

Taken together, these results indicate the very real potential to employ second generation wood pellets in coal handling systems with relatively minor modifications.

As noted previously, mills 3B and 3C were physically modified for this test by removing their discharge skirts. The mill coordination curves (fuel-air curves) for these mills were also slightly modified to increase the mass flow of primary air at the lower end of the mill range.

This was done based on white pellet pneumatic transport experience acquired during previous testing within the OPG fleet. The revised fuel-air curve is shown in Figure 3 on page 26.

The biomass mills were also benchmarked with air flow (no fuel) to identify the expected mill differential and motor power for an empty pulverizer. This was done to yield a clean state set of data that was used to determine when the mill was clean following a shutdown.

Following a normal mill shutdown, the cleaning air flow and temperatures were set similar to the values noted here. When the pulverizer differential and motor power approached their “clean” states, the mill was considered to be empty.

With the boiler stable and operating on coal, the first modified biomass mill was placed into service on September 10, 2013. The classifier vanes were set to fully open (position “0”) for this initial test.

The mill differential and motor power both stabilized very quickly during this first test. However, it was also immediately apparent that the mill was rapidly rejecting fuel through the throat. It was necessary to increase primary air flow to eliminate the fuel rejection issue. The mill did stabilize at the minimum feeder setting with an air flow slightly above the modified coordination curve value.

This reject problem can be attributed to the use of the original mill throat that was designed for operation with air temperatures in the range of about 300ºC. Operation with colder air flows for biomass firing (initially about 120ºC) resulted in a significant increase in air density and therefore a similar drop in air volume and throat velocity. This mismatch was corrected after the initial testing by modifying the free flow area of the throat to better suit the new operating regime.

Mill Classifier Discharge Skirt (prior to removal)

Parametric testing was conducted on the modified mills, including the variation of fuel flow, air flow and classifier position. A good operational fit was determined at the mid-range classifier vane position 5 (about a 45º vane angle) and the mills was tested throughout the necessary load range.

Mill differential and motor power were both seen to increase significantly when the classifier was adjusted from the fully open position to the mid-range “five” setting used here. This observation was contrary to our experience with pilot scale mill testing, where the variation of the static classifier blades did not result in any significant impact on the solid particle performance.

Also, following the closing of the classifier vanes, mill rejects were observed to increase again. A large positive primary air flow bias was necessary to address this issue, again reinforcing the need to modify the mill throats for this service. The increased demand for air flow is very likely linked to an increase in fuel recirculation with the tightening of the classifier vanes. This was not observed at pilot scale but is certainly the expected trend with coal operation.

The industry is still learning what level of fineness is needed for efficient combustion performance when firing biomass fuels in suspension. Experience from existing white wood pellets conversion projects seems to indicate that about 90 percent passing 1000 microns is suitable for well-designed burner systems. Achieving that level of fuel fineness with white wood pellets in modified coal pulverizers is another matter entirely.

The results presented here for advanced wood pellet grinding routinely meet and exceed this threshold for fuel fineness. The commissioning team has instead used the fraction passing 500 microns as the main indicator of mill performance in this regard. This approach is somewhat more conservative and the values observed in this regime display more variation to operational adjustments.

For a fixed classifier position, pulverized fuel fineness is rather flat over the load range of the mill. This level of fineness has been found to result in well-defied, stable flames and a very bright and clear furnace environment. The fuel feed rates tested here are actually higher than the required flows for full boiler MCR with one mill out of service.

Advanced Wood Pellets at Thunder Bay GS (August 2013)

Power consumption is seen to increase with the fuel feed rate and it is likely that motor power will be the load limiting factor in most cases. As indicated above, in the Thunder Bay scenario, the equipment was able to operate at high fuel flows, sufficient to supply properly sized fuel at the original nameplate capacity of the unit.

The specific mill power — expressed as kW/Mg — is also observed to remain relatively constant over the load range of the mill. The full scale results in this area (about 11-14 kW/Mg) are significantly lower than the values determined from the pilot scale rig (22-40 kW/Mg). This favourable variance is thought to be linked to the much higher dead weight and applied force in a full scale pulverizer.

The primary air temperature entering the mills was limited to about 120ºC for the initial proof of concept testing. A lower temperature was considered to reduce the risk of mill fires with the highly volatile fuel. The specific figure of 120ºC was selected based on OPG experience with white wood pellet milling.

This temperature was found to be suitable during testing, as indicated by the effective drying of the pellet fuel down to the inherent moisture level. Mill performance, pneumatic transport and combustion were all also found to be acceptable.

A key objective of the second week of field testing was to evaluate the ability to start the unit on the advanced wood pellet fuel. Up to this point, the unit had been started on coal, with the biomass fuel replacing coal on a unit at about half load. For the first biomass start, the oil ignitors and warm up guns were put into service per the standard protocols, to heat up the furnace and raise boiler pressure.

Mill 3B was placed into service using the same parameters and procedures that had proved successful in the first week of the test burn. However, almost immediately, very high carbon monoxide (CO) levels were observed that did not decrease until well after the mill was taken out of service. Several other attempt were made to start the mill with the same negative results. These unsuccessful attempts included the use of a lower fuel feed rate in an attempt to improve pulverized fuel fineness.

Observations of the grinding roll deflection during these failed starts revealed that the fuel bed was very thin. Another mill start was executed that increased the fuel flow in an attempt to establish a thicker fuel bed. This effort did prove successful as this optimized mill configuration resulted in a minor CO emission spike that reduced to a level of less than 20 ppmv in several minutes. The adjustment of the low load mill fuel flow along with monitoring furnace temperatures prior to the admission of solid fuel has resulted in trouble-free unit starts since this initial issue was encountered.

The initial test series also evaluated the capability of the unit with respect to steam temperatures, as the use of wood pellets in a boiler designed for lignite coal has been shown to result in issues in that area. The steam treated pellets used at Thunder Bay were initially tested at unit loads of about 25 percent and 50 percent MCR. In this range, steam temperatures were indeed slightly lower than for operation on the baseline coal. There was no real attempt to increase temperatures during the initial test burn. The burner elevation tilts were in the zero to 10-degree (up) range.

In general, combustion performance during the initial trial with advanced wood pellets was excellent. Stable, bright flames were easily established without oil support, resulting in a clear furnace environment. No burner system adjustments were required.

This image also indicates significant flame detachment on the level “C” burner (the lower of the B/C firing combination). This appears to be directly attributable to the use of higher than optimal primary air flow to control the fuel reject rate during early testing. As discussed earlier, higher primary air flow was required as the pulverizer throats were not yet properly modified.

Observations of the fireside conditions also indicated very little fuel falling into the bottom ash system.

A project scope for a full unit conversion was developed to fully leverage the performance benefits observed in the initial field trials. The project had a clear focus on safety, with adjustments or modifications throughout the fuel handling system. The major items are listed here.

The capital expenditure for the conversion of Thunder Bay Unit 3 was approximately $3M (Canadian dollars), resulting in a specific capital investment of less than $25/kW (Canadian dollars, net capacity). This value obviously compares well with similar metrics for white wood pellet conversions, where the specific capital expenditure is typically in the range of $500-800/kW.

The unit entered into commercial service via an energy supply agreement in January 2015. Thunder Bay Unit 3 became the first coal fired unit (worldwide) to be fully converted to employ advanced wood pellets as the primary fuel.

The parameters of the energy supply agreement require that Thunder Bay Unit 3 is capable of a net electrical output of 135 MWe. This load is readily achievable and in practice, the unit can easily operate at the original full coal-fired nameplate capacity of 153 MWe (net).

As a peaking unit, Thunder Bay Unit 3 typically operates in a grid support role, as the electricity demand in Northwestern Ontario remains low following the economic downturn of 2009. The limited run-time on the unit has not allowed for major optimization activities but valuable operating experience has been gained in a number of areas.

In the case of Thunder Bay, the advanced wood pellets need to be stored outdoors for extended periods of time. This involves exposure during all seasons including winters than can be very cold and result in significant precipitation. The peaking role of Thunder Bay contributes to the duration of storage but a key component is the abbreviated shipping season on the Great Lakes. Thunder Bay receives deliveries of advanced wood pellets via ship on Lake Superior but boat transport is forbidden in the December to April timeframe each year due to frozen conditions on the Great Lakes and the difficulty in obtaining insurance for water-borne cargo.

As a result, Thunder Bay stores advanced wood pellets outdoors, without the benefit of cover, for periods up to one year in duration. Analysis is conducted on samples from all working piles on a monthly basis to track pellet degradation and moisture uptake.

Pellet durability is seen to remain quite high, indicating that the mechanical strength of the pellets has not been significantly compromised by the period in storage. With the exception of an outlier for May 2016, the fines values were also found to be very good.

It is important to note that the total moisture values tend to track with seasonal precipitation and also include the impact of evaporation in the hotter or drier months of the year. The relatively high values for total moisture also tend to support the use of a rather severe lab scale weathering treatment.

Indeed, moisture values above 20 percent are never observed for lab scale soaking durations of only 48 hours.

The true saturated total moisture level of these particular advanced wood pellets is only realized after a soaking duration of 1 week (168 hours).

These samples were collected from the surface of the November 2015 pile and therefore represent a somewhat conservative view of the integrity of the entire fuel volume. OPG and others have conducted testing to confirm that there exists a “sheltering” effect in piles of advanced wood pellets.

In this regard, the outer surface layer that is directly exposed to the elements can be expected to degrade faster than the bulk of the pile that forms the core of the stored volume.

OPG has tested this phenomenon at full scale with the Thunder Bay working piles and has observed a distinct benefit in the durability and fines metrics for pellets that are at least 20-50 cm below the surface of an outdoor pile.

Dust control when handling the advanced wood pellets has been excellent. The operations staff have gained sufficient experience in this area such that several of the dust suppression stations can be idle when handling fuel that is already wet naturally as a consequence of outdoor storage. Fuel spillage is also well controlled but when housekeeping is necessary, field experience has shown that dry vacuuming is the preferred method. Washing down galleries with water – the practice on coal — has been found to result in problems handling the runoff water in the powerhouse.

Although the pellets handle very well in a dry or wet state, two of the modifications to the process did create a challenge in the first months of commercial service.

First, the installation of the rotary air locks upstream of the feeders has created a bottleneck where relatively small pieces of debris can plug the air lock channels, effectively tripping that mill.

Second, the frozen coal crackers were taken out of service to avoid unnecessary attrition on the pellets. Wet pellets that subsequently freeze are capable of forming large clumps, about 20 cm in diameter.

These large frozen masses are capable of plugging fuel feeders or the rotary air locks. These issues have been addressed at site by increased diligence during fuel reclaim and by the installation of grating upstream of the coal crackers.

This latter simple retrofit effectively screens large debris that could be problematic for the downstream equipment.

Operation with pellets that are high in total moisture content has been observed to impact both the thermal and mechanical capacity of the pulverizers. The parameters that impact the drying performance of advanced wood pellets are basically those for coal use, dominated by the inlet and outlet primary air temperatures and the air/fuel ratio. When analyzing the drying process in detail, users are advised to employ a specific heat relationship developed for wood as this does vary somewhat from the coal baseline. Field results to date indicate that the wood pellets tend to behave like a bituminous coal, in that they are capable of evaporating all of their surface moisture content, given suitable conditions.

The firing of pellet fuels with total moisture levels above 20 percent allowed the operations staff the opportunity to evaluate the normal means of dealing with wet fuel in a pulverizer.

Modest increase in primary air flow were found to be effective and the range of higher air/fuel ratios tested did not have a negative impact on combustion. Higher primary air inlet temperatures have also been employed – up to 160ºC — significantly higher than the 120ºC baseline used in the initial test burn program. The use of higher air temperatures has had the expected effect and in normal operation, has not been observed to represent a safety issue. However, during the first trials with elevated air temperatures, several pulverizer fires were encountered, all on a single mill (mill 3D). The cause of these fires was traced to an improperly adjusted pyrite sweep that was allowing fuel to accumulate in the mill windbox beneath the table. Correction of this issue resulted in trouble-free operation, highlighting the critical need to ensure that high volatile matter wood fuel is not allowed to accumulate in this area.

Operation with wet fuel also has a negative impact on the capacity of the mill. Experience has demonstrated that the stability and capacity of the pulverizer is quite sensitive to the ability of the mill to expeditiously grind fuel and transfer it to the outlet pipes with little or no fuel recirculation. The higher density of wet fuel particles and a lower mill body velocities caused by higher density (colder) air can both contribute to fuel accumulating in the grinding zone. This in turn has been seen to limit the capacity of the mill via excessive mill differential pressure or motor current. Promoting better drying and a higher mill outlet temperature is key to maintaining the capacity and stability of the mills with wet fuel.

As discussed earlier, even with the removal of the discharge skirt, the position of the classifier vanes has been confirmed to impact the performance of the mills when firing advanced wood pellets. Opening the classifier vanes slightly — from position 4 to 3 — has resulted in additional margin for the mills when handling wet fuel and has had only a minor impact on pulverized fuel fineness.

Perhaps the single largest physical change from the first test burn to the full converted unit is the modification of the mill throat area to increase velocity in this region while using air flows closer to the design case. Recovering some of the additional primary air flow has reduced the burner tip velocities, promoting better flame attachment while maintaining the same bright, clear furnace environment.

The final operational configuration of the mills employs a slightly higher primary air flow curve and a classifier vane position that is more open. Both of these adjustments provide additional margin when operating with wet fuel and give the operators flexibility at maximum unit loads to operate with either four or five mills in service.

Operations heats up the furnace and raises pressure with auxiliary light oil firing as normal. The current practice also includes firing until the furnace exit gas temperature crosses a threshold value determined during commissioning. An infrared temperature meter is employed for this purpose. The combination of a sufficiently hot furnace environment and the use of lower mill elevations for the first biomass mills in service has resulted in excellent results.

As a peaking unit, Thunder Bay Unit 3 is usually required to operate at the low end of its load range. However, when called upon to operate at full load, the unit has demonstrated the capability to easily generate at the full (original coal) nameplate capacity. Table 8 includes a set of operating data from one such campaign.

This run was conducted in September 2015 and consumed fuel from deliveries in the fall of 2014. The pellets were handled and fired without incident but the long period of outdoor storage did increase the moisture content, resulting in an artificially high total fuel flow. Also, the relatively elevated value for excess oxygen is a requirement of the stations’ environmental permit when firing biomass fuel. This value has been demonstrated to be conservative and could be lowered in practice.

The main steam temperature is seen to increase to the design value. However, the hot reheat figure is still almost 20ºC below the optimal value. Maximizing the reheat temperature is not necessarily a priority for operations but would likely require optimization of the burner tilts angles and perhaps some level of fuel biasing in the furnace.

Opacity performance on the advanced wood pellet fuel has always been excellent. The hot-side precipitators installed as original equipment at Thunder Bay handle the flyash very well. It is assumed that reducing the excess air level at full load to a more typical value would offer the opportunity to reduce opacity even further.

Aside from the previously discussed issues with CO production on the initial biomass mill start, CO emissions have also been good. The combination of good pulverized fuel fineness and high volatile matter results in excellent combustion conditions, including a clear furnace and the elimination of secondary combustion and burning carryover.

In November 2016, the first source test (stack emissions and dispersion modelling) was completed on Thunder Bay Unit 3 firing advanced wood pellet fuel. This testing was conducted in accordance with the station environmental permit and also serves to provide the first complete accounting of the emissions profile for the converted unit.

The results from the source test program were used to demonstrate compliance with the pollutant criteria set forth in the stations’ Amended Environmental Compliance Approval (ECA, air permit).

Particulate emissions have been a significant concern in industry, particularly when older technology, such as wood waste combustors (beehive burners) are employed. The combination of a suspension combustion system with a utility scale precipitator has resulted in excellent TSP results at Thunder Bay.

As discussed earlier, CO emissions have been very good on the converted unit, typically in the range of 20 ppm at stable load. Auditing very low pollutant levels has become problematic, as the required accuracy and repeatability is a challenge for the stack monitors when operating with low pollutant levels. CO production was actually increased during this run to allow for a better check of the continuous emission monitoring system.

The results for NOx are the closest to the ECA limit but still comfortably below the threshold. It should be noted that this level of performance has been achieved without the benefit of any in-furnace or post-combustion NOx controls. Experience with biomass units in Europe has indicated that OFA and SCR systems are effective NOx reduction techniques with wood firing, so several paths towards further progress in this area are available.

Mercury, sulphur and PM 10/2.5 emissions are all also current issues in the power sector. The nature of the feedstock (clean forestry) yields the expected results for both mercury and SO2 emissions. The fine particulate results are also very favourable, the product of a low ash fuel, good combustion and particulate collection performance with the existing equipment.

The majority of the test samples for mercury were actually below the detection limit of the approved method. Only those results above the detection limit were used in the calculations and reported.

The program also included triplicate testing of metals, polycyclic aromatic hydrocarbons (PAH), dioxins and furans. Similar to the case with mercury emissions, many of these trains were analyzed and found to be below the detection limits for their respective methods. Only those results above the detection limit were used and reported, resulting in a conservative analysis. All of the results in this area are well below existing regulatory limits.

The stack emissions profile offers many advantages over the baseline coal case, especially in the areas of acid gas emissions and mercury. However, the main driver for displacing coal with biomass fuels is the net reduction in greenhouse gas (GHG) emissions. A full accounting of the GHG footprint of a station firing a given fuel requires a life cycle approach, where the impacts of upstream fuel production and delivery are considered, in addition to the base GHG intensity during operation. OPG supported a peer-reviewed study of the life cycle analysis for the use of advanced wood pellets at Thunder Bay to define and understand the issues at hand.

Both pellet cases assume that the biogenic CO2 emissions during combustion are eventually balanced by uptake in the forests and therefore, do not contribute to atmospheric GHG levels. In this respect, only non-CO2 GHG (CH4, N2O) are considered at the point of use, resulting in a minor contribution to the GHG footprint. All cases include fuel production, fuel transportation to site and point of use combustion.

The conversion of Thunder Bay Unit 3 from coal to advanced wood pellet firing is the first such project in the world. This project confirms the ability to execute a low capital cost conversion project by leveraging the unique properties of these new second generation biomass fuels. The Thunder Bay case study has demonstrated the potential for utilities to execute similar projects to repurpose existing coal assets using advanced biomass fuels as a means of increasing their portfolio of dispatchable, renewable power.

Les Marshall is Senior Technical Officer Ontario Power Generation Canada