Al Ferrer, VP Power Consulting, George Keller, Senior Consulting Engineer, Burns and Roe; and
David Perri, Director Corporate Engineering, Barry Durham,Corporate Environmental Manager, Bohdan Dackow, Corporate I&C Engineer, USPowerGen.
The New York State Department of Environmental Conservation (DEC) has mandated that the Astoria Generating Co. (AGC) install variable frequency drives (VFDs) on the circulating water pumps on three of the operating units at th Astoria Generating Station in Astoria, Queens, New York. The objective is to reduce the annual flow of river water through the plant and, as a result, reduce impingement and entrainment (IM & E) of aquatic organisms and minimize environmental impacts.
Challenges result from reducing the cooling flow that must be addressed in the overall system design. Specifically, reducing the cooling flow reduces the facility efficiency and increases the temperature rise (discharge vs. intake temperature) through the facility. These effects can conflict with requirements within the facility’s State of New York State Pollutant Discharge Elimination System (SPDES) discharge permit. The SPDES permit sets a maximum temperature rise between intake and discharge and an absolute upper limit for discharge temperature that must be continuously adhered to. In addition, there needs to be economic considerations that limit the adverse impacts to the facility’s heat rate. These are examples of the many constraints that the system control logic had to deal with.
The VFD system’s purpose is to reduce the flow of circulating water through the plant by reducing the speed of the circulating water pumps. This needs to be done within the constraints of permit limits and acceptable performance penalties. The acceptable performance penalty is determined based on plant economics and environmental requirements. This parameter is one that can vary seasonally or be based on other operating variables and, as such, is an operator settable variable.
The performance penalty control is driven by an innovative condenser pressure setpoint program. This logic is implemented in the unit distributed control system (DCS) using two lookup tables: the condenser performance curves and the bogey turbine heat rate curves. This allows USPowerGen to balance IM&E vs. heat rate penalty costs in real time.
Our goal was to automate and optimize the control scheme. Our understanding is that most VFD systems run in manual mode, where an operator sets the speed of the pump, which does not optimize the system. We believe this system will be capable of running in full automatic (optimization) mode.
Astoria Generating Station
The Astoria Generating Station is a 1,280 MW fuel oil and natural gas plant in Astoria, Queens (in New York City), bounded by the East River and Luyster Creek. The plant has four operating steam turbine units. The plant is part of a larger complex that is comprised of around 300 acres, most of which is used for various utility purposes including other power generation.
The U.S. Environmental Protection Agency (EPA) and the New York State Department of Environmental Conservation (NYSDEC) have oversight of Astoria’s once-through circulating cooling water system (CWS). The EPA has initiated the 316(b) regulations which require a systematic review, design and implementation of modifications to intake structures of existing power plants as well as operational modifications to reduce organism impingement and entrainment. The NYSDEC has jurisdiction to grant SPDES permits for waste water discharge including cooling water. As the 316(b) studies progressed for the Astoria station, USPowerGen submitted an application to renew its SPDES permit to the NYSDEC in September 2005.
In addition to the SPDES permit renewal and the 316(b) work, Astoria had also been granted an Article X certificate to repower the facility. The repowering plans contemplated a significant reduction in cooling water use from the river since the current cooling system would have been replaced by a closed cycle cooling/wet cooling tower system which is considered to be 316(b) compliant. As a result of the repowering being delayed and the initiation of the 316(b) rules, the NYSDEC was required to renew the SPDES permit to reflect the continued operation of the existing Astoria plant within the compliance requirements of the new regulations.
As part of the renewal, the NYSDEC required the facility to install variable speed pumps (VSP) on Units 30, 40 and 50’s CWS to reduce facility water throughput. The Astoria SPDES permit was issued in March 2007 with the VSP requirement.
Two primary objectives are associated with the work to install VSPs: compliance with the facility SPDES permit and integrating the required upgrades into the overall facility’s 316(b) compliance planning.
The overall SPDES program and 316(b) compliance program is estimated to take approximately five years to complete since the construction work will need to be coordinated with major scheduled outages to minimize unit down time. The scope of work beyond installing VSPs will only become clear once the effects of this work have been determined by aquatic sampling programs.
Once the requirement to install VSPs became effective, several steps were required to determine how to approach the technical challenge of upgrading the facility. The primary tasks were as follows:
- Review existing equipmentDetermine the limitations of the equipment to operate at reduced speed as well as understand the state of repair of ancillary components in the system.
- Engineer the modificationsDetermine what components are needed and optimize the design.
- Develop variable speed pump operating methodologyDetermine how to operate the multiple units.
- Procure the components for the VSP installation
- Construct the modifications during scheduled unit outages.
USPowerGen spent a significant amount of time evaluating the existing equipment and developing an engineering strategy. This was completed with the assistance of an architect/engineer, pump experts and various vendors.
The primary findings from this effort were:
- The pumps could operate in a range of approximately 54 percent to 100 percent
- The existing motors could be modified for VFD operation
- It was cost effective and feasible to maintain the existing 440V level for the pump motors in lieu of replacing the motors and feeds with a higher voltage
- A condenser vacuum priming system was recommended to maintain the condenser water boxes in a flooded state
- The overall operating philosophy was complex because the inlet and outlet temperature for the plant have permit limits, however the units are being controlled individually but discharging to a common discharge point. The NYSDEC did not provide any guidance on the operating concept.
The initial operating philosophy did not adequately address the potential for a unit’s heat rate to be significantly affected by reducing the condenser water flow, nor did it allow the operator to manage flow rates to each unit to optimize the entire plant.
The control system design clearly needed to be further refined. The task was determined to be beyond the capabilities of any likely system installer. As such, a controls expert was engaged to assist USPowerGen in developing a system that met the objective of maintaining compliance with the facilities’ absolute temperature discharge limit as well as the maximum plant delta T limits allowing operator set points to define the “allowable heat rate penalty.” This provides flexibility to balance environmental performance with heat rate considerations.
Developing the Controls Logic
Burns and Roe was hired by USPowerGen to develop a control strategy and the logic to minimize circulating water pumps speed without undue penalty to the heat rate and without exceeding discharge temperature and differential temperature permit limits. Burns and Roe also was to review the initial AE reports, existing plant permits and operational records and conduct discussions with USPowerGen environmental and operations personnel to ensure the control strategy completely addressed all issues.
The initial approach to circulating water pumps speed control optimization was deemed too complicated and too expensive to implement. It also did not address plant performance concerns. Thus we started our search for the least complicated logic that would still work and comply with all requirements.
A fair number of “easy” decisions were made. These are, however, easy only for the very experienced and can be overlooked by others.
Failure Modes: It was agreed that the control system failure mode should reset circulating water pump(s) to full speed. Not doing so could result in occasional non-compliance with discharge temperature/diff temperature permits, which is not acceptable.
For the same reason, it was decided that the speed control signal should be reverse acting with a limited range: 4 mA = 100 percent (full) speed, 20 mA = 50 percent speed. If this was not done, a damaged wire, loss of power supply or blown fuse would force the pump to zero speed, leading to non-compliance with discharge temperature, differential temperature permit limits and/or a unit trip.
Ramp Rate Limit: It was also agreed that the Ramp Rate Limit would reside in the VFD controller as well as in the DCS. This would prevent possible damage to the pump should something happen to the current output or wire integrity of the control loop.
A number of more difficult decisions were encountered. For example, since the station has a common water intake and discharge, there was considerable debate and apprehension on the subject of Master/Masterless design for discharge temperature control. A common Master would assure permit compliance and prevent possible interactions between three independent unit temperature controllers.
There were two negatives to this approach, however. First, it would create a single failure point. (Masterless design is more reliable.) Second, it would be awkward to implement due to a multi-year installation and commissioning schedule for the three units. (A fourth unit at the station connects to the same discharge tunnel but is not required to have a VFD.)
In this case we had to be innovative in the system design. Process control theory tells us that multiple gain-only pressure reducing controllers working on the same header can interact. They “fight each other,” which is visible as load swapping. The solution is to make sure that the gains are not the same. In such case, interaction stops. The theory is silent on the parallel operation of multiple temperature controllers. From our tuning experience, parallel loops slower than pressure-reducing controllers do not interact with gain-only (no reset or integral action) controllers. With strong reset action they always interact. Since the water exit temperature discharge is such a slow loop-vs.-pump speed, if we made the temperature controllers gain-only, chances of interaction are either negligible or at least not observable in real time. So a Masterless design would be quite do-able.
Next, we considered an interaction between the override temperature controller and the cooling pump speed controller. The cooling pump speed controller always tries to reduce pump speed up to a point. The override temperature controller will block it and even increase pump speed if the station discharge temperature approaches the limit. This dual function provides opportunity for control instability.
From experience we knew the only good way to set up an override control is by using a function generator F(x) and a signal selector. One can shape the F(x) so that the temperature controller stays out of the way of the VFD controller until an action must be taken. The last segment of F(x) sets up the trigger point and amplitude of the action.
Results of the evaluation concluded that each unit VFD logic should be independent and reside in its unit DCS. Common intake and discharge temperature measurements will be propagated (via a current loop) to all three logics. Two engineer tunable function generators will provide high discharge temperature and “high diff” temperature override control. Lastly, it will be easy to bias any unit so one can run with more or less flow from the others.
Following debate it was decided that condenser pressure is the best process variable to control the pumps. Condenser pressure is a function of the water temperature in the condenser tubes. Small changes in the condenser pressure are linear with the small changes in the average water temperature. Small changes in the average water temperature are inversely proportional to the cooling water flow and, thus, the pump speed.
Pump speed changes quickly resulting in condenser pressure changes which is good for loop stability. Condenser pressure is a robust signal that is easily measured. A decrease in pump speed of 27.5 percent will result in a roughly 1” to 2” Hg increase in the condenser pressure and roughly 4 percent increase in the heat rate. Again, to avoid possible interactions, we use a proportional-only controller to increase VFDs speed should the condenser pressure increase vs. its setpoint and vice versa.
The condenser pressure setpoint was also debated. There were suggestions of an on-line optimization scheme to generate the setpoint, but that was out of the budget for this project. A suggestion was also made for a simple fixed set point based on the design heat rate. USPowerGen indicated that a fixed set point was unacceptable and a better solution was required.
In the end, this challenge resulted in a much better solution. The clever part is the condenser pressure setpoint program. This logic is developed in the DCS using two lookup tables: the condenser performance curves and the bogey turbine heat rate curves. This allows the Astoria station to set the acceptable unit performance and balance IM&E vs. heat rate penalty in real time. It is this innovative approach that made this project noteworthy. Here are the details:
Using the condenser performance curves, the intake water temperature and megawatt output, the unit logic calculates the theoretical condenser pressure at full pump speed.
Using the bogey heat rate curves, the condenser pressure at full pump speed and the megawatt output, the unit logic then calculates the heat rate at full pump speed. Using the heat rate curves, the heat rate at full pump speed and the acceptable heat rate penalty in Btu/kwhr (operator adjusted), the unit logic calculates the heat rate at reduced pump speed. Using the heat rate at reduced pump speed, the unit logic then calculates the desired condenser pressure setpoint.
The speed of this iterative loop could be as slow as 30 to 60 seconds. The default value of the condenser pressure setpoint (after a reload or an iterative loop failure) should be 1.5” Hg. An auto/manual setpoint station with operator settable Hi-Lo limits is provided just in case. It is anticipated that the pump speed will be changed at a rate not exceeding roughly 10 percent a minute.
The result: knowing the Btu penalty, Astoria station personnel can balance reduction in water flow (which is theoretically proportional to aquatic effects) vs air emissions vs unit economics in real time.
Which is the true condenser pressure? Actually it is not that easy to tell. Therefore multiple pressure measurements were used, at least two on each shell. The measurements should ideally be located one to three feet above the tubes. To measure true condenser pressure and avoid velocity head bias, a basket tip in accordance with PTC 12.2, Fig. 4.2 should be used. Sensing tubing should be 1/2” SS minimum, continuously sloping down to the tap, with the transmitter mounted above the tap. Otherwise condensate build-up will distort the measurement.
To get the best accuracy of intake and discharge temperature, we had to use three wire 100 Ohm wound platinum RTD’s, which are sensitive to vibration. Thermocouples are much more durable but 10 times less accurate in our temperature range. Because of the permits, the accuracy of intake and discharge temperature measurements was not negotiable. Next, we had to worry about protecting the RTDs at a reasonable cost.
Putting RTDs in a 316 SS sheath with 0.25” OD was a good start. However, the 316 SS sheath cannot go directly into practically sea water; it will pit very quickly. We had to use thermowells for protection.
Monel is the perfect material for a thermowell in sea water service. It will not pit and does not allow bio growth. Bio growth would increase the time constant of the temperature measurement and thus would make discharge temperature control more difficult if not impossible. However, we needed thermowells over 20 feet long (to allow for a tidal variation) and they were just too expensive in pure monel.
The solution was different for the discharge vs intake measurement. On the discharge we had a strong current, over 5 feet per second (fps). Here we used Sch. 160 316 SS thermowell. It was economical and the current would keep it clean (no build-up) with minimal pitting. On the intake we had no current, 0 fps. Not a good application for a 316 SS thermowell. The solution was a two-foot-long monel thermowell with a 23-foot-long FRP holding pipe. The holding pipe was screwed into the monel T/W and made water tight at the factory. Since there is no current and vibration, this installation works.
The concern with waterbox level was what might happen in the unlikely event of a priming system failure. The answer: water level will not drop immediately. If it does drop, it will impair heat transfer and cause tube vibration. Since our units have titanium condenser tubes, no damage was expected in the short run. Impaired heat transfer very quickly results in condenser pressure increase. This will cause the unit logic to increase the pump speed until the heat transfer and thus the water level are restored. The problem is self-correcting due to the choice of the process variable. The whole water level drop transient should not last more than 30 to 60 seconds.
Therefore, we did not add the waterbox level measurement to the VFD logic to force the pump speed up directly from the water level signal. However, we did add a digital input to the VFD logic, that, when closed, will force pump speed to 100 percent. This input will be for future use, if proven necessary.
We decided that each unit VFDs will operate in parallel, all the time. No sequencing will be used due to the configuration of piping from the pumps into the condensers.
We settled on 54 percent as the pump minimum speed. We selected this based on a study to determine what the limiting factors were on the pump that would drive the minimum speed. Items such as bearings, condenser tube velocity and motor fan cooling were reviewed. Ultimately the minimum tube velocity was the limiting factor to maintain cleanliness of the condenser tubes.
Once the control scheme was developed, it had to be implemented. Of greatest importance was to maintain the highest possible level of reliability for the circulating water system (CWS). Along with reliability there many other design decisions that needed to be evaluated and acted upon.
One decision that needed to be made was whether to reuse the existing 440V motors, rewind the motors for VFD operation or upgrade the system to new 4160V motors. Since the existing motors were quite large at 700HP, there was concern that a reliable 440V VFD system could not be procured and would require us to change to a 4160V system at a hefty cost. USPowerGen determined that systems of this size were available and had been successfully implemented. Once this was established we were confident that keeping the existing 440V system was technically feasible and cost beneficial.
The next decision was to decide if the existing 700HP motors needed to be rewound for VFD operation. Rather than making the decision ourselves we decided to define a required reliability/warranty period for the motors and let the bidders decide what was required. After the bidding process it was clear that it was necessary to rewind the motors with VFD rated insulation to ensure long term system reliability. Since this design was going to be used for a total of seven motors (six in service with one spare) it made sense to require a full load type test of the rewound motor and the new VFD in the factory prior to final delivery to the site. Of particular concern was the possibility of insufficient cooling of the motor at reduced speeds since the cooling fans for the motor windings were intergral to the motor rotor.
A final area of concern for reliability was the VFD itself. Prior to the need for varying the speed the system was simple being comprised of a 440V circuit breaker, cables, and a 440V, 700HP induction motor directly coupled to the pump and had been operating reliably for over 35 years. Since the addition of all the components of a large VFD system reduces the overall system reliability, it was decided to include an automatic bypass with the VFD. This bypass system would automatically switch from the VFD mode to “across the line” operation in the event that any component in the VFD system was to fail.
However, just adding a bypass alone does not improve reliability if it is not done properly. The design of this bypass system was done by the VFD supplier with feedback from the owner to ensure it was a robust failsafe design. Once the design was finalized and the system was built, a thorough owner-witnessed factory acceptance test was performed. As with most custom system designs, problems were identified with the control scheme, that necessitated some changes. All of the problems were successfully resolved in the factory prior to shipment.
By spending the additional funds to have and attend a detailed factory acceptance test we were able to identify these problems and correct them quickly with the appropriate resources. Waiting until critical equipment is installed in the field to perform functional testing can be a risky proposition. When issues were identified in the factory, all of the equipment and expertise were available with the time needed to implement well developed solutions. Because of this approach we feel that we received a solid and reliable system.
More Power Engineering Issue Articles Power Engineerng Issue Archives View Power Generation Articles on PennEnergy.com