1 Inlet heating system
ESEC installed two MHI 501G gas turbine, the site weather in winter time come be down to -32 degree C, when the humidity is bigger than 70% degree C, at the rime of the inlet IGC will form the forest from inlet camera, but from the second stage of the inlet compressor which is high chance to form the icing which due to the IGV will throttle the air and decrease the dew point to cause the moisture in the air will form the icing to attach it to the surface of the blade to damage the blade if the icing come off from the blade surface to cause the damage.
Figure 1 winter time site conditions
So the MHI install a anti-icing system which will operation if the temperature drop to 10 degree C and gas turbine below 70% load the anti-icing will start operation to let the 6 stage compressed hot air will send to inlet of air house to increase the air to prevent the icing form on the IGV, in the meaning time there is inlet heating system installed if the temperature below -15 degree to use it to heating the air back to -15 degree C, and MHI not require the inlet heating and anti-icing operation if the gas turbine operation at base load which means IGV open up to 100% .
So the MHI install a anti-icing system which will operation if the temperature drop to 10 degree C and gas turbine below 70% load the anti-icing will start operation to let the 6 stage compressed hot air will send to inlet of air house to increase the air to prevent the icing form on the IGV, in the meaning time there is inlet heating system installed if the temperature below -15 degree to use it to heating the air back to -15 degree C, and MHI not require the inlet heating and anti-icing operation if the gas turbine operation at base load which means IGV open up to 100% .
After the gas turbine commissioning, it is found the IGV will be icing ether at based load lf the humidity is greater than 70% this cause the big challenge hot how to use the inlet heating system to increase inlet air temperature, increase the inlet air temperature plant output will decrease. If not increase inlet air temperature the potential icing will possible damage the gas turbine blade .
Figure 2 Inlet heating system
Figure 3 CTG air inlet
3 Challenge.
Gas turbines will produce more power when the inlet air temperature is colder due to the increased density of the air, but what many combined cycle/ gas turbine operators don’t realize is that it’s the moisture in the air that the OEM’s are the most concerned about. When temperatures get cold, moisture in the inlet air will quickly form ice on the combustion turbines IGV’s, bell-mouth and inlet support struts which has the potential to sluff off and impact the rotating blades, Mitsubishi’s original control philosophy for inlet heating failed to address this relationship and was independent of relative humidity. Calgary in winter can down to -32 degree C, plant site humidity change frequency due to many factor, CT1 was tripped at 2016 due to extremely cold weather to freeze IGV, in the cold weather especially at high humidity condition to heating inlet air to avoid safety concerns is necessary, but if too much inlet heating which will decrease gas turbine output, to find a best way to control inlet air heating input to optimize plant output is challenge.
4 Solution.
Plant staffs discuss with the MHPS to upgrade the control logical and build icing curve to automatically control the inlet heating input to optimize the plant out as follow measurement successfully implemented
- Install a camera at inlet of the Buell month to monitor the icing condition
- Relocate the relative humidity sensors to after inlet heating coils, two humidity sensors to response the humidity change before and after inlet heating.
- The idea of temperatures and relativity humidity having an icing correlation is called an icing curve The icing curves will utilize the relative humidity from the RHT transmitters downstream of the heating coils to provide an accurate picture of the air being supplied to the CT.
- Create a graphic for each GT that plots the temp/RH before the heating coil and the temp/RH after the heating coil. The plots will be on top of an icing curve developed with MHPS and operations experience and be focused on base load operation (IGV 100%). The graphic may need to vary based on IGV position ,Intent is to have a red x that moved with actual inlet conditions
- Set up inlet heating steam control to auto control heating input, to optimize the plant output
- Gas turbine output forecast tool add the humidity impact to response the inlet heating to change the plant output hence to meet grid dispatch in different weather condition
- Upgrade anti-icing control philosophy is to determine the best time anti-icing in-service, The revised philosophy properly accounts for ambient temperature, relative humidity and IGV angle which will protect the gas turbine away form icing damage
Fig 4 icing curve for the CTI, red dot is show the operation results
5 Test Process
5.1 parallel flow test
Over the past two winters the CT inlet preheating coils have been susceptible to frost buildup during ambient icing conditions while the units are operating at baseload. Once the icing has begun, priority manual scraping is required to remove the buildup. Derates have be incurred as a result.
Due to the counterflow design of the glycol coils, fin density and multiple tube passes, limited heat is transferred to the leading edge of the coils. With the high inlet air velocity of humid air, this cold face leads to frost buildup.
Site has investigated multiple options to resolve this issue. This has included engaging the original supplier (Mitsubishi), filter house designer (Braden) and heating coil manufacturer (Marlo DRS). Mitsubishi has summarized their recommendations in a letter which has been added to the MOC folder.
One proposed solution is to reconfigure the glycol piping to result in a parallel flow heat exchanger. Supplying the hot coil to the lead edge of the heat exchanger would increase the likelihood that the frosting scenario can be avoided.
With direction from ESEC , SiFi has proposed 3 options to achieve the desired arrangement…
#1 – Swap at 6″ headers
#2 – Swap at main 8″ headers
#3 – Swap at individual coils
Each arrangement will be designed with valving to allow for returning the flow path to the original counterflow design.
Refer to the MOC folder for details on proposed layouts, pros, cons and estimated costs.
It is proposed to proceed with one of the three options on CT2 during Q4 2017. This will serve as a trial for viability allowing for a full implementation or reassessment in 2018.
This change will mitigate ice build up on the glycol coils and associated derates. The original OEM, Braden, has stated it is not possible to successfully model the impact of the flow change. Although it will reduce the likelihood of icing building up on the leading face of the radiators, it cannot be guaranteed to what extent.
Reverse flow through the heating coils will also increase the potential for air locking. This will have to be monitored operationally.
3″ glycol piping is to be modified at the connections to the upper-most coil on the north side of CT2’s inlet filter house. The piping configuration will allow for glycol to be redirected to operate the individual heating coil in either a parallel flow or counter flow configuration.
P&ID markup and preliminary models for the piping configuration are stored in the MOC folder. SiFi has been engaged to perform the detailed engineering and fabrication for this temporary modification. Formal drawing updates will occur if this configuration is to be made permanent.
The trial is targeted to occur between Q4 2018 and Q1 2019. This temporary modification will serve as a trial for viability of reversing flow. The system will either be further modified or returned to its original configuration in the summer of 2019.
The intent of this change is to mitigate ice build up on the glycol coils and associated derates. Over the past winters the CT inlet preheating coils have been susceptible to frost buildup during ambient icing conditions while the units are operating at baseload. Once the icing has begun, priority manual scraping is required to remove the buildup. Derates have be incurred as a result.
The original OEM, Braden, has stated it is not possible to successfully model the impact of the flow change. Although it will reduce the likelihood of icing building up on the leading face of the radiators, it cannot be guaranteed to what extent. Braden has also warned about potential iisues with air locking of the system as the coils are not designed for parallel flow.
Based on modeling and theoretical calculations, AAF concluded that the swap “would not provide the appropriate amount of energy to keep the fins leading edge out of the icing zone.”
Reverse flow through the heating coils will also increase the potential for air locking. This will have to be monitored operationally.
In a parallel flow configuration, the system’s overall heating capacity/efficiency will be reduced.
5.2 Control logic update
The control logic for the flow valves does not recognize whether or not a pump is running. As such, when the glycol pumps shut off, the two flow control valves (1/2-IAP-FV-78036 & 1/2-IAP-FV-78029) go to 100% open trying to achieve their flow setpoints -approx. 230t/hr to the coil and 175t/hr bypass. If the system is started with these two control valves in auto and sitting at 100% the pump will either trip, or the flow will go bad quality rejecting the control valves to manual. One suggestion would be to add logic recognizing that the glycol pumps are off and sending the flow control valves to a pre-determined “pump start” position. The bypass FV should sit at around 50% and the coil FV should be 0%. This would allow the pump to start with all fv in auto while drawing less amps and ensuring proper minimum flow control. At 50% on the bypass FV and 0% on the coil FV the pump flow is around ~185t/hr. Once the pumps have started and everything has settled out the normal logic can take over with the glycol coil supply control valve (1/2-IAP-FV-78029) slowly ramping to its desired setpoint of 230t/hr and the bypass pinching in to maintain 175t/hr minimum pump flow (eventually closing once coil exceeds 175t/hr). This can cause an imbalance in the system, valves rejecting to manual, over amp the pumps, and lock the system to the point that valves have to be forced. By adding a transfer block between the PID and MA station, we can set the valve position at the desired value while the pumps are off and allow them to return to normal operation shortly after a pump is running. We will also place a transfer block on the set point so that the set point for flow through the coils will ramp from 0 to 235(current SP) at a rate of 2/sec. This will give the system time to balance itself.
5.3 system update
Replace existing contractor style 231 expansion joints at the suction and discharge of IAP glycol pumps with contractor HP. Initially, all discharge expansion joints will be upgraded as they have been problematic. The suction expansion joints will be replaced as required in the future.
IAP glycol pump discharge expansion joint 1/2IAP-EJ-002/004 (existing Item# 100244045)
IAP glycol pump suction expansion joint 1/2IAP-EJ-001/003
Design Conditions:
Fluid: 60% propylene glycol, 40% water
Design Temperature: 108°C
Design Pressure: 1138 kPa
Connections: ANSI 150# FF
Dimensions: 4” nominal diameter x 6” length (discharge); 6” nominal diameter x 6” length (suction)
Other: designed for full vacuum
Notable changes:
– Rubber material to change from neoprene to EPDM. This will provide improved performance at elevated temperatures.
– Teguflex expansion joint contains 316SS vacuum ring
– Teguflex expansion joints are only rated for 30mm of compression (versus 35mm for Proco model), however this is sufficient for the application. Teguflex’s allowable extension, lateral and angular movements all exceed those of Proco
– 4″- diameter Tegulflex expansion joint has spring rates of 65 kg/cm, 85 kg/cm and 65 kg/cm for compression extension and lateral, respectively. 44, 63 and 73 kg/cm for the respective Proco spring rates. These differences are expected to have a negligible impact on the system operation.
– 6″- diameter Tegulflex expansion joint has spring rates of 75 kg/cm, 100 kg/cm and 75 kg/cm for compression extension and lateral, respectively. 68, 99 and 126 kg/cm for the respective Proco spring rates. These differences are expected to have a negligible impact on the system operation.
Item records to be created to allow for future ordering and replacement as required. Item# 100303508 for the discharge expansion joint. Assets to be created for expansion joints, as necessary.
Drawing to be obtained from contractor to supersede GG-SH-00-EMP-ACW-DD-0001-001. Alternatively, the contractor drawing will be revised to note tag numbers that no longer apply. Existing expansion joints have been subject to failure during startup of the IAP glycol system, specially on the pump discharge. Due to the expansion joint design, they are sensitive to over torquing and are easily damaged when attempting to seal any leaks.
6 Results.
The site humidity is change by hours, the anti-icing curve provide obviously indication the best operation zone, if the red dot is out of the optimization zone operator will decrease inlet heating input immediately to increase plant out, if the red dot is move in side inside of icing zone, the alarm will come, and combined with inlet camera results, operator can increase inlet heating to avoid the risk of icing.
In the past with the ambient temperature at -10C, with the prior established methodology given to the plant by previous management, the plant could only ever produce ~810 MW’s. However, with the new RH curve inlet heating methodology recently implemented, it is found the plant be able to generate ~828 MW’s when the RH is below 80%. 15MW for just less than a 1/2 a year at $35 per MW it translates to over $2million dollars per year.