BLOG: Power Perspectives

[#1]

Some emergency and standby power applications require the ability to start up and accept electrical loads in less than 10 seconds. If a project has a power restoration time requirement, it is highly recommended that a system level Start Time Analysis (STA) is completed. The STA should look at every aspect of the critical power path, from electrical system component selection to component performance (including settings, time delays, and inherent component processing delays) to site conditions that may impact system performance. The following list offers recommendations to achieve faster generator set starting.

Configuration

• Avoid unnecessary generator oversizing when configuring generator sets that require a fast start time. Oversized and high voltage generators have a larger rotational inertia than the standard low voltage generator, and will increase generator set start times.
• In a paralleled generator set system, ensure the critical loads are less than the capacity of the smallest generator set and consider energizing all critical loads as soon as the first generator set breaker closes to the bus.
• Consider using a remote radiator with electric driven fans to decrease the inertia during starting on the generator set package, particularly when using an oversized or high voltage generator. Using oversized or high voltage generators in conjunction with engine driven fans may increase start times into an unacceptable range.

 
Controls

• Ensure package electronic controllers are updated to the latest software version.
• Engines equipped with electronic governors should be properly tuned for the site and application. This includes reviewing the ramp rate, and adjusting from the default value to improve start time. Note: increasing the ramp rate will result in increased black smoke on start-up.
• Packages equipped with a Digital Voltage Regulator (DVR) should be properly tuned for the site and application. This includes decreasing the "soft start" feature in the DVR, which will decrease the time it takes the generator to build rated voltage.

 
Electric Starters & Batteries

• Use heavy duty electric starters with fully charged batteries.
• For generator sets with only a single starter, add a second starter.
• Ensure the electric starter and battery set is able to crank the engine above 110 RPM for ten seconds.
• Use a redundant/back-up battery charger to ensure fully charged batteries.

 
Air Starters

• Depending on engine model, air starters may increase or decrease cranking speeds and thus affects the overall start time.
• For air start systems, maintain adequate air pressure to crank engine; size air tanks and related lines to crank engine above 110 RPM for at least ten seconds.

 
Temperature Control

• Use jacket water heaters to maintain jacket water temperature at 32°C (90°F) minimum.
• Supply combustion air at 21°C (70°F) minimum.
• Supply battery heaters if ambient temperatures are below 0°C (32°F).

 
Fuel/Fluids

• Have a readily available supply of clean fuel.
• Ensure fuel filters are clean, and there is no air in the fuel system due to leaks.
• Use continuous engine oil pre-lubrication, if available with the engine model.

 
The parameters listed above cannot guarantee starting in a certain number of seconds, but will certainly improve start time.
 

• Have you ever performed a Start Time Analysis?
• If the analysis resulted in a start time greater than 10 seconds, how did you resolve the issue?
• When you exercise your genset (weekly/monthly), is start-up time one of the parameters you measure?
• Are there additional measures that you know of that can be taken to improve start time?

  
Please share your insights by posting below.

[#2]

As systems and generators get bigger, this seems to come up as a problem more often.  I rarely get involved on the front end, and unfortunately sizing isn't always controlled by what is techncially best these days.

Newer engines with electronic controls make it easy to evaluate the crank, ramp and overall time to rated speed and volts using the datalog features in CAT ET, setup the logging to look at the IC input, the SMMS output, actual engine speed, actual average volts and frequency, setup datalog for fastest log rate and do a test.

Older engines require some external test equipment and depending on system can sometimes be difficult to get all the info in a format to compare.

Electric fans can be a real help, as long as fan controls have a delay to allow the critical loads to come up first and then bring fans on after things settle, I have some systems where we don't start the fans for 3 minutes to meet all the site needs.  On larger systems with remote radiators consider multiple smaller fans, staging them as needed, or a VFD controlled fan.

On ramp rate and/or start fuel limit, in some cases highest ramp rate or start fuel limit may not provide best startup speed.  So before just cranking it up all the way do some testing.  In 3500 mechanical engines a properly adjusted start fuel limit on the 2301A had a better start time than with the start fuel limit not active.  Have also found similar results in some electronic engines, of course with the electronic engines easier to measure with ET.

On digital AVR's, a similar issue to governors, the fastest ramp rate may not give you best possible time to breaker close or transfer, especially if it results in a large overshoot.  Each system will have it's own dynamic response, so careful attention during startup is important.

Batteries and starting systems can have a huge impact, and unfortunately what is supplied standard from CAT sometimes isn't up to the task.  During crank minimize voltage drop, this may require increasing battery cable size, additional batteries, and minimzing connections, things like removing the standard battery disconnect switch on larger engines can have a measurable impact.  A DMM with recording feature, like a Fluke 87 is the best tool to help qualifying the condition and ability of the starting system.  A worksheet is helpful for two reasons, one it helps the technician do the investigation in a logical manner, and two provides documentation for future reference.

On temperature control, in some applications we have found that engine coolant temps in the 125 degree F range improved overall start and load better than settings less than 100 degrees F.  One of biggest impacts on cranking speed is lube oil temp, a warm block and cold pan can still result in slow crank speeds, meaning a slower start.  Lube oil heaters are expensive, but depending on situation can be a feasible solution.

Over the years fuel system piping and plumbing problems have accounted for a large number of "slow starts" after units have been in service awhile.  During install and testing, units run a fairly large amount of time compared to being in regular service and sitting for days or weeks at a time between starts.

Prelube pumps may help if the engine controls have a speed ramp delay based on oil pressure, have done comparisons with prelube equipped engines with and without prelube and never found an improvement in actual start time because of running a prelube, however engine service life is certainly improved.  Just make sure prelube cycyle doesn't have negative impact, likely flooding over turbo seals.  Engines equipped for continuos prelube have bypass circuits to prevent oil from going above top of block.

In some extreme cases we've used air impingment on turbo's for large engines, but due to complexity and service issues I would try to avoid if possible.

In my experience a system approach is best, many times been called to a site with problems meeting required start times being blamed on the engine, only to find a combination of problems to real root cause.  And every once in a while I come across a system that just won't meet the target, the best field adjusting and tuning can't overcome a poor design or poorly done specification.

My two cents, Mike L.

[#3]

With 1 genset, you guys are right on track. You have some excellent suggestions.

If you have more than 1 genset, there is a way advanced controls can help. The process has a lot of names, but basically the synchronization occurs as the excitation ramps up. Here is the process: (re-printed from EGSA's - www.egsa.org - upcoming book on power generation)

Close Before Excitation (CBE) or Delayed Excitation Mode. This is an old mode of starting a generator that has found new life with the need to meet a requirement of NFPA 110: restoring Life Safety Power in 10 seconds. When the total load in this category exceeds the capacity of one generator, CBE can be used to start multiple generators in less than 10 seconds. This is the sequence of CBE:

1. A start command is sent to all generators at the same time. Some generator protections are disabled (ANSI 27, 59, 81U, 24, for example) while others 50/51, 32, 81O are left functional.
2. A short time after the gensets’ RPM exceeds the crank RPM, the output breaker is closed on all gensets. Note: the excitation is not allowed to automatically start at this time like on standard applications.
3. Once the first genset speed reaches about 70% of rated AND enough generators are running to support the load:

            i.      any engines still on crank or not up to speed have their breakers opened and they are locked out of the process until step 7.

            ii.      the excitation is started on all running generators

            iii.      excitation is gradually ramped up, typically 1-2 seconds from zero to rated volts.

1. Gensets that did not start within this time may be brought on-line via standard synchronization practices later.
2. As the excitation increases, the generators will be pulled into synchronization. Tests have shown that the current typically stays below 20% of rated, where a standard synchronization of even 15 degrees out can lead to 80% or more of rated current flow. This also reduces the electrical and mechanical stress on the generator windings, circuit breakers, transformers, couplings, and the engine itself.
3. Once voltage and frequency reach 95-98% of rated values, the generator protection is enabled and the system load breaker is closed and the system is on line.
4. Gensets that did not start in time – step 3 – may now be added to the power system with the standard synchronization process.

 Again, this is not a new process. It can work in some form with Gen-to-utility installations, but it works really well with multiple-gen installations. It is working today at Uof Miami Hospital and other places around the world.

Feel free to email me with any questions.

Steve Evans

see@deif.com

[#4]

Happy day!

What are your thoughts on the necessity of prelube?

Enjoy the day with sunny smiles to share :-)

[#5]

 For Prime application, no one cares it, but it is an imporatant issiue for Standby gensets.

Nowadays,  multi-gensets are used in standby system and oversize genset would be selected because the load type.

I think the fast starting time is not fittable for a big isolated multi-gensets system. The designer should conside more in reliability than saving time.

And I still have a question for 10s starting time. That is how you guarentee the air temp above 21 degree C if you are in cold weather.

If you seal the inlet and outlet duct before start, some devices have to be added and the reliability is affeccted.

Additional, when I add pre-lube system, the cost is expensive and maintenanec works are increased. 

[#6]

Nice discussion.

[#7]

Wow, dead field paralleling, I had hoped that was long dead based on previous experience (all bad).  We went thru a round of systems trying that in the late 80's and early 90's, a couple of local consulting engineers thought it was the ticket.  We could make the systems work, but keeping them working presented many problems.

Given that control technology has improved greatly since then, maybe it's worth trying again, however while we could get many of these type systems work to get thru initial testing and acceptance, ongoing and long term operation was problematic, and a couple of times resulted in some pretty messy failures. Not one of the systems we installed during that time period stayed in operation as initially installed, they all were converted back to "standard" paralleling systems and most are still in operation today.

I would love to see the engineering community maybe take a different tack, how about getting rid of this ridiculous and archaic specification.  Let's face it, if you have a critical process today, it is backed up by one or multiple UPS systems, UPS system reliability is far better than it was and frankly a 1 second outage is as big a problem as a 10 second outage in most critical applications today.  Trying to make bigger and bigger systems start, parallel and accept large loads in 10 seconds really is hard on equipment, usually doesn't work "as planned" a couple of years after installation, and is prone to problems due to lack of maintenance, adjustments, and technical competence to properly support some of these systems.

On prelube, I'm a fan of prelube for the right reasons. I never found a case where it made an engine start better except with older hydramechanical governors, especially on D349 and D348 gensets, and some 300 series units. On modern electronic engines it likely has no real benefit to improving start.

Prelube is a benefit for one primary reason, at engine crank, instead of the engine rotating 20 to 90 revolutions before oil is delivered to the main bearings on a cold and dry engine (after sitting for 24 hours), the oil gets under the bearing in as little as 10 revolutions, greatly reducing bearing wear occurring from the start cycle.  Many years ago some people did quite a bit of testing at Lafayette on 3500 and 3600 engines to really figure out just what was happening with (and without) prelube.  Of course the guy that lead that (and many other) fact finding mission has retired and last I heard hanging doors for a living.

CAT prelube systems are sometimes not well done, mainly poor service life from pumps and pressure switches, and they can't seem to develop an overall strategy for dealing with it.  The best systems can do intermittent prelube while in standby mode, and a timed prelube cycle during engine start. In some cases post lube is also desirable.  A flexible control system is best to allow ability to meet customer and site needs.

Mike L.

[#8]

Spoiler

Response to mikel:

Unfortunately, the National Electrical Code (NEC) often drives the 10 second requirement:  NEC Article 700 (emergency systems) identifies equipment that MUST come back within 10 seconds of loss of Utility power.  For most occupancies, this is egress lighting, exit signs, etc.  You can skin that cat using battery powered lighting (comes back almost instantly) and/or battery powered equipment....  but because the BATTERY has an expected life of about 5 years, you're faced with replacing the battery every 5 years.  For a fluorescent battery ballast, that cost is about $350 per ballast, PLUS labor to replace, PLUS degraded reliability until it DOES get replaced... And that's for EVERY fixture in the facility....  so, run the numbers but for an owner occupied building, you might not want to go this way.

Similarly, NEC article 701 (legally required standby) describes loads that MUST come back within 60 seconds of loss of Utility power.  This becomes important (and REQUIRES the installation of a generator) in a few occupancies (Ihigh rise building, large assembly occupancies) or for certain types of equipment (some types of fire pumps, mechanically ventilated smoke evacuation systems, some types of elevators). 

NEC Article 702 (optional standby loads) describes OTHER loads that don't really have to come back at all, but you might WANT to make them come back automatically (or manually)...

In hospitals, NEC Article 517 (health care facilities) requires a three-branch system, with similar 10 second, 60 second and automatic-or-manual transfer requirements...  except that in health care facilities, the third branch contains equipment that is not exactly "optional", but may also contain other "optional" loads.  (NFPA 99 and state hospital licensing requirements may further define what equipment must be placed on which branch of the hospital emergency generator system).

In data centers, nearly ALL of the load would be considered by the NEC to be "optional standby" (NEC 702), so a 10 second start time is a bit overkill.....  unless....  unless any NEC 700 load is placed on the generation system...

One thing to keep in mind about NEC 700 load: Not only must it come back within 10 seconds, but it must also NOT share any of the conduit raceway system as other "non-NEC 700" circuits....  including sharing the transfer switch with any "non-NEC 700" loads... 

Another thing to keep in mind about NEC 700 load:  For most facilities, the load can often be covered by fairly small KW, even for a fairly large facility....  Even where "legally required standby" loads may range in the 1000's of KW, the "emergency" load is often under 100 KW...

This suggests an interesting solution when dealing with multiple generators:  A "pony" generator in the sub-100KW range dedicated to EMERGENCY (NEC 700) loads, which starts in under 10 seconds serving a completely separate bus...  Don't parallel this pony generator with ANYTHING.  I'm thinking it shouldn't be any kind of special starting technique to get a sub-100 KW gen to start in under 10 seconds (WAY under 10 seconds)....  My car engine (equivalent of about 80 KW of generation) starts in under 2 seconds on a COLD morning with a so-so battery...  For the remainder of the load, provide your standard "big bank" of paralleled generators, but arrange to allow start in 60 seconds ... or possibly longer if there are no NEC 701 loads being served from the "big bank".    After all, why should 80KW of "10 second" start-time load dictate a 10 second start for a bank of multiple 2 MW generators???

And...  the pony generator wouldn't even need to be the same fuel source as the 702 generator (for example, if the local AHU has requirements regarding "on-site storage" for back-up fuel, or if there are emissions considerations).  Or, for the REALLY bold:  a TURBINE for the 702 loads (say in the range of 10 MINUTES or more to come up, stabilize and get ready to accept load)...

Concerned about run-time mismatch between the NEC 700 generator and the NEC 702 generator?  Consider a secondary transfer switch downstream of the 702 generator, ahead to the pony generator transfer switch....  arrange transfer switches such that should the 702 generator every stop (or fail to start, or run dry, or whatever), that the "pony" generator transfer switch will re-start the pony generator...  that SHOULD satisfy the local code authority that you have ASSURED backup for NEC 700 loads, regardless weather the "normal" side of the 700 transfer switch being fed by Utility or being fed by the 702 generator.

The point being:  Consider matching the back-up generation technology - including start time - to the exact requirement of the bus being served, there may be advantages to backing up several busses from more than one generator.  Consider a single bus backup source ONLY when economics or physical size constraints dictate.

thanks.

jt

[#9]

Good Comments

[#10]

Sorry to hear you had problems with Dead Field Paralleling in the past. It works great at the Univ of Miami hospitals with three 2.5MW units.

We also have used it for many years on ships. Marine applications can really be quite complex with gensets of different sizes, fixed speed gens, and shore power connections. They can't afford to have an outage when maneuvering in harbors.

Were the problems you experienced, Mike, related to control system failures or mechanical engine failures? New integrated control systems are much less likely to experience failures and if they do, they safely shutdown the genset to prevent major failures.

Another consideration: three 500KW gens are usually cheaper than one 1.5MW unit. So if you max demand is 1.5MW, you can use four or five 500KW gens for much less than two 1.5MW units and get fast starting, increased reliability, fuel economy, and many other benefits, all with advanced control systems. I am talking about controllers that cost less than $5K...peanuts compared to another genset.

One last comment: dead field paralleling is not a new technology. It was used to bring large steam turbine-generators on line generations ago.  

Best regards,

Steve

[#11]

I was directly involved with 6 systems that attempted to use dead field paralleling, 3 two engine systems, 2 three engine systems and 1 four engine systems. The rating were between 500ekw and 1750 ekw. All the systems I worked on were same set parallel systems, meaning same size and type units.  None of the systems experienced any prime mover problems as a result of the dead field parallel system.  All of the faults were related to multiple breaker trips, blown surge suppressors, failed diodes, three failed AVR's, two failed excitors and one failed rotor.  All systems were specified by the same consulting engineer using two different switchgear companies.  As I said above, these were late 80's and into early 90's when this work was done.  Critical standby systems, at least in our area, still relied heavily on relay logic and "old school" thinking.  Every system installed was commissioned as dead field parallel and all passed initial acceptance testing.  Earliest failure was on the first monthly test after commissioning, resulting in a failed AVR.  Just about every system was retrofitted back to "standard" phase and voltage match synchronizing within 1 year of original install.  The last system, which is when the rotor failure occured, was the last straw, our sales department would not bid on a dead field system after that and as far as I know none went into our area again.

I'm sure with newer controls and protective devices, likely a system could be done if someone was willing to spend the time and effort (oh yeah, and money) on it.  However with the state of the standby market these days, the continuing drive to reduce cost, footprint and installation time, plus the reduced number of run hours to test/adjust it if you live in a strict emissions area, doesn't really seem to make it feasible unless either one of the genset manufacturers or major switchgear companies would do the R&D needed to build comfort and recommend and support those kinds of systems.

By the way, not a single failure of any of the generator failures encountered with the dead field parallel systems were paid by CAT warranty, we ate every one.  I have a friend who works for Cummins and he has experience with two older systems that also initially used dead field paralleling, both of those were also retrofitted. Same type issues but I don't have details.

I'm fully aware dead field parallel has been arond for a while, I've heard a large number of hydro's use it, however don't know of a hydro system with a 10 second start requirement.

My 2.5 cents worth, Mike L.

[#12]

Please find under below youtube link an impressive example regarding static paralleling/ unexcited startup/ Russian synchronization, dead field synchronization.

You can see what is possible with an integrated, high sophisticated Woodward easYgen-3500 P/N 8440-1934 genset controller without using additional external logic/PLC and hocus pocus/mumbojumbo. The Screen in the middle of the panel is more or less just used to show the complete application on a TFT screen.

http://www.youtube.com/watch?v=6zqqAdUSqpU

Data center in Ankara/ Turkey, Customer Turkey Telecom

    4 x 1MW Caterpillar diesel generator sets, engines preheated

• All for sets are in standby, datacenter supplied by local grid

• Mainsfailure, black out (18.sec during the movie)

• Mains failure delay time configured to 7 seconds (18.-25. Sec during the movie)

• After 7 seconds all 4 generator sets  equipped with Woodward easYgen-3500 controllers are starting together without activated  excitation (AVR switched off) with a closed Generator Circuit Breaker GCB

• All 4 diesel generator sets are running on nominal speed (29 sec during the movie)

• After, all 4 AVR’s are switched on simultaniously and group breaker (as well served from easYgen-3500 controller) is closed after all 4 sets are running on nominal speed. (31.sec full load supplied by diesel generator sets)

• 4 x 1 MW diesel generator sets paralleled within 6 seconds (from diesel generator start command to full load supply)

[#13]

Our Close Before Excitation or what some call Dead Field Paralleling works great. We have been using it for years on gensets of various manufacturers. Like all complex control systems, it must be installed and started correctly. We see our competition doing it now, too, so there must be a demand for it.

Prelube and other engine operating systems must, of course, be operational/ready. With the three 2.5MW Kohler/MTU gensets running at U of Miami, they start in 6.78 seconds repeatedly. We can even set the start delay timer to 2 seconds and NOT start for short outages and STILL make it under 10 seconds.  The 10 second rule will not go away. It will only get tighter as more and more systems depend more and more on electric power. I don't suggest you push on that rope!

Mike, I have seen many of your posts and they are excellent. You say your experience with failures is from the 1980's or 90's. When I was doing large AC and DC drives in the 1980's and 90's, we had many more failures then they do today. I am sure you will agree that controls and electronics have changed dramatically since then...remember the bag phone?  Maybe it's time to re-visit this control scheme with modern controls, exciters, and power electronics? 

I am not saying you are wrong: I have seen your replies to others and highly respect your work. Today we are working with CAT more then ever in the past. We have our controls on literally hundreds of CAT gensets worldwide in Marine and Land applications. We know controls: it is all we do!  CBE works and does not lead to component -- mechanical or electrical -- failures.  

Best regards,

Steve