Hydro Optimization Team Meeting – McNary, January 20, 2000

Introductions (Exhibits 1 and 2)

Tom Murphy reviewed the agenda. The team reviewed the original charter, clarifying that the HOT primarily recommends actions rather than implementing, although the focus seems to be shifting. The charter was approved as written.

Is there interest in identifying formal membership? The team would prefer to update the distribution list, keep consensus voting, and allow open attendance.

Recap of Reclamation operations discussion from 9/11/06

Tom reviewed the action items from the previous day’s discussion of operations with Reclamation:

· Look at RAS and see how it affects optimization and recommend changes (Lou Tauber-lead, Kent Anderson, Tom Murphy, operations, TBL, Bent Mouritsen)

· Look at on-off costs for big units, comparing condensing vs. turning on-off (general rule is they stay on in condensing mode if they are needed in the next 12 hours since restart reliability in ten minutes is poor); model wear and tear costs for field breakers, governors, exciters, generators, and turbines. Consider the effects on transmission system reliability. Toby Steves-lead with Denver design.

· Denver will also look at on-off costs for smaller units – need some criteria to help determine when to change unit commitment. WaterView currently uses costs of $300-$400 per on-off cycle – this figure needs to be confirmed, or a new estimate should be developed. However, it could be problematic collecting data. Toby-lead, with Richard Nelson, Tom Murphy, plus operations and electrical project folks to determine better estimates. Reclamation may have some R&D funds in the Power Resources office to help in the analysis – which would be applicable to many plants.

· BPA is also looking at improving forecasts for hour-ahead and future-hour project generation requirements. Fran Halpin is working to get the mid-Columbia plants to schedule the bias.

· Need to rectify curve differences between NRTO and WaterView, looking at performance data relative to model tests. Toby and John Brooks will create curve updates for optimizer.

· Kent Anderson will write a new Coulee standard operating order for operators emphasizing certain operations – it will be an interim procedure and will be evaluated using feedback from operators.

In general, efficiency is higher when there are fewer small units on-line before a large unit is started. For flexibility reasons Coulee operators like to have most of the 18 smaller units on-line before they start a big one, but that leads to the larger efficiency losses.

BPA schedulers will try to reduce plant loading losses based on specific real-time plant operations by sending out setpoints that avoid areas of high losses.

· After Kent writes the new operating order, he will make a presentation to the BPA Duty Schedulers. The presentation will show dollar-saving benefits by operating per the recommendations of the applications.

Feed Forward AGC (Exhibit 3)

Tom Murphy reviewed examples of how the feed forward AGC will send signals to the operators to help them make decisions on when to turn units on or off. Bart McManus of the Transmission Business Line (TBL) proposes to send signals based on allocations, i.e., sending block requests to projects using response factors. The goal is to reduce losses at the projects for within-hour generation changes.

The new signal holds requested generation flatter for a longer period of time, rather than sending dynamic 4 second changes. The AGC system will calculate a MW change based on the plant response factor.

BPA will use a short-term simulator for Columbia Vista (CV) to improve forecasts out 1 to 12 hours by showing trends and targets However, the difficulty in making these forecasts will change depending on time of day, current river operations, and the role of the specific plant in the system.

Overview of Mid-Columbia River projects and hourly coordination (Exhibit 4)

Fran Halpin reviewed concerns raised by owners, operators, stakeholders, and fisheries from the operations of the 7 mid-Columbia plants (5 non-federal projects, plus CHJ and GCL). He explained how the generating capacity varies widely among the plants, but the hydraulic capacity is more even. Coordination among the projects is intended to improve overall generation efficiency for the system as a whole by maximizing head with Target Forebay Elevations called Anticipatory Spill Protection Elevation (ASPE).

The issue for the non-federal plants is that they stand to lose some unintended flexibility (load following and regulation) they have enjoyed as demand and constraints increase on federal generation. Another current issue being addressed in the hourly coordination agreement is the equitable distribution of benefits resulting from coordination.

BPA believes that the mid-Cs should also be doing more load factoring themselves and should be following the ramping of their thermal projects during the day rather than relying on the federal system. This would be facilitated by hourly pre-schedules.

BPA is also working to resolve issues related to bias implementation on a 4 second interval, which can cause unnecessary unit starts. There is a significant cost to the federal system from these unnecessary stops and starts to support the within hour bias changes. Under current agreement, the federal bias is only supposed to provide bias changes for hydro balancing every 20 minutes for balancing hydraulics across the federal and non-federal plants. There are several events which make it allowable for the bias to change more frequently, but BPA is working toward to reducing FCRPS exposure to within-hour bias changes.

NRTO can show the dollar costs for the loss of efficiency to the federal projects, plus the mechanical costs to the units of start/stops. The members of the Hourly Coordination Operating Group have agreed that the current algorithm which has been driving the operation for decades doesn’t work very well at times. The Hourly Coordination Technical Committee is looking to redo the algorithm so that it will be more forward looking by including non-federal load forecasting. New candidate algorithms should be developed and testing initiated this year.

During the three year timeframe when work is going on in the CHJ tailrace for installation of the spillway deflector, the bias must be scheduled hourly for safety. The FCRPS is using this as an opportunity to test impacts on the hourly coordination of different bias separation timeframes.

There has been some disagreement over the definition of hydraulic coordination and where the burden of unit start/stops, load following, and regulation should fall. BPA contends that hydraulic coordination involves planning, and that no hydraulic adjustment (bias) should occur until a change in load can be seen to be persistent and not simply a transient load anomaly. Many of the non-federal parties believe that anything which results in forebay movement (even in the four-second timeframe) constitutes a need for hydraulic adjustment (bias). BPA is pushing for these adjustments to be made in anticipation of future needs and be limited to one change per hour under normal conditions.

BPA would like to see more policy representation from Reclamation at Hourly Coordination policy group meetings particularly at the monthly Operating Group meetings.

WaterView vs. NRTO (Exhibit 5)

Toby Steves compared screen shots from the two optimization systems. He was interested in determining why there was a discrepancy, and found that it was due to efficiency bounds. The greatest efficiency benefits at a plant come from having the right number of units on. The systems have to find a balance - the efficiency band cannot be too narrow because that results in too many changes in unit operations.

To make the two systems correlate better, Toby suggested some alternatives. He could force WaterView to perform commitment changes if the benefit is smaller, or match WaterView’s “Up Margin” calculation in NRTO. There are pros and cons for each approach.

There was a brief discussion on how to communicate with operators: the systems need to provide feedback on dollars and have WaterView give suggested actions that the operator can compare with his or her own decisions and understanding of specific system constraints. In general, WaterView only provides a snapshot that doesn’t work during ramp times.

WaterView will add feed forward – but BPA needs to provide information at least 20 minutes ahead if the system will propose adding new units to mix. WaterView needs as wide a window as possible, with as much accuracy as possible on the direction that generation is going, as well as information on specific amounts of generation that will be needed.

Air injection effect on HRS cavitation meter on G24 (Exhibit 6)

Toby Steves reported that during a preliminary test, acoustic cavitation almost went away with air injection into the turbine environment, using a condensing air system. He still plans a formal test of the effect on units 24 and 19 which have acoustic cavitation meters installed, when he can get a unit outage and look at the response. He is interested in seeing if the bounds of the rough zone can be reduced, which gives more flexibility and will decrease cavitation repair.

Turbine testing

Toby expects to receive a report on cavitation from the turbine testing.

Corps minutes

The Corps attendees adopted the 5/23/06 minutes with an addition from Richard Nelson.

T1 Index test box (Exhibit 7)

Lee Sheldon described tests of a successful proof of concept of a new way to develop cam curves. He summarized the conclusions, lessons learned, and recommendations.

Rod Wittinger commented that HDC has dropped the recommendation to purchase additional index test boxes from the contractor as a result of contracting difficulties. Instead they intend to develop functioning prototypes with in-house efforts, using codes owned by the government. They will need to develop an automatic flushing system for the Winter-Kennedys.

Richard Nelson concluded that an independent ITB should reside in the digital governor, so software needs to be able to be integrated into the generic governor controller.

Dan Ramirez will replace Rod Wittinger and lead a design team to address a number of index testing issues.

T2 - Type 2 for powerhouse

Ed Miska reported that prototype software was successfully installed at CHJ and BON, and will be completed on time. However, before ICCP is installed at other sites, more testing is needed to increase confidence, and HDC also needs to test feed forward at these projects to verify that it’s stable, and track and analyze data results before using them.

HDC has developed a requirements document for T2, added new features to integrate with GDACS, and included new functions for economic dispatch and unit commitment. T2 will suggest start/stop operations and show a dollar value – which is advisory only for operators. Economic dispatch will run on a periodic basis, and only suggest changes depending on certain threshold criteria – the system will not suggest frequent on-off operations but will provide conservative suggestions for the operator.

To work correctly, this system will need BPA feed-forward data, so those forecast features will not be immediately turned on and those functions won’t be tested this fiscal year. There will be an incremental addition of functions for the operators. BPA will focus on developing feed forward accuracy in order to be available for T2: Tom Murphy will push TBL to develop to meet the ICCP deployment date. Eventually T2 will replace most NRTO functions at the projects – Tom reports that most operators are now comfortable with NRTO, so they should be ok with T2 on GDACS screens.

The unit loading feature of T2 uses a 5 minute rule to run efficiency program every 5 minutes, with suggestions only if the benefit is above a certain threshold. The new program will adjust unit loading at 10 minutes after the hour with 20 minute signals, resulting in two within-hour changes.

Can T2 be tested sufficiently with existing resources? Richard Nelson noted that contractual acceptance testing is underway now and ongoing debugging will continue with implementation.

The next version of T2 needs to include: individual ICCP functionality; feed forward testing; individual unit gross head values; capability to use individual unit performance tables; and a control data communication exchange with AGC.

HDC mechanical engineers will check derivatives and algorithms to see if there could be a document and configuration control issue, especially for performance control curves that need to be updated in GDACS. HDC engineers have the responsibility for updating the cam curves. Richard noted that the NWW portion of 3-D cams are installed in GDACS, but he doesn’t know who updates the Portland 3-D cams. Generally, district approval is needed before sending the new curves to fish agencies for BiOp implementation, so each district needs to clearly identify and prepare the path for approval.

SCHEDULE:

Complete T2 evaluation and test at CHJ/BON – thru 12/31/06

Deploy ICCP – FY 07 (needs to stay on schedule)

Deploy T2 Version 1.0 (software installation) at remainder of GDACS projects at same time ICCP is installed in FY 07 (use Version 1 to look for bugs, and hold training)

Install feed forward system (Tom will work with TBL to develop a schedule)

Follow-up actions:

· HDC will confirm whether the GDACS tables need to be updated for T2. (flow table update)

· HDC will report whether T2 works in the field and whether the operators are happy with it.

· Tom Murphy will push TBL to develop to meet the ICCP deployment date (Tom-lead; Ed for design and configuration; with Kent Anderson, Steve Davis, Dick Hammer, Scott Bennett, and Glen Smith).

· Clarify path for approval of 3-D cam curve information and performance and flow information.

NRTO status

Tom Murphy reported that the benefits report is being updated to reflect actual outages, and then it will be sent out to the projects on a weekly basis. It appears that virtually every plant is operating with reduced and acceptable efficiency losses, but Tom cautioned that operators need to consider on-off costs when making their decisions. Overall, operators have cut unit commitment losses in half, saving about 30 MW at $300k-$400k/year per MW.

Absolute flow (Exhibit 8)

Lee Sheldon described the results of flow measurement tests on LWG unit 4. His group has concluded that determination of absolute flow is difficult in short intake Kaplans, particularly with fish screens in place. T2 needs absolute flow in order to obtain an accurate optimization solution. However, HDC still doesn’t know what level of accuracy is needed. A consistent bias error would still allow for an accurate T2 optimum solution, but problems arise when there are different families of units at a plant.

Lee compared results from absolute measurement methods such as acoustic scintillation and acoustic time of flight (Accusonics) with the Winter-Kennedy piezometer system for relative measurement. It appears that measuring time of flight without screens is the most accurate method.

Lee would like to issue a new solicitation to ask for new methods of flow measurement, or to find out if there are improvements in existing methods (e.g., changes in calibration), or to borrow methods from other industries, or reduce the cost of present methods. He would take the proposed technology to a lab with a calibrated flume to determine whether intake conditions (flow turbulence, air, etc.) affect the accuracy. He would use the lab to determine which method is more accurate (rather than expensive testing in a field environment). Lee is hoping that there are new methods available to accommodate different levels of variability, and wants an error band with the costs of implementation per unit factored in so that a choice can be made.

When preparing a solicitation, it is important to consider what HDC will get as a final system (e.g., how many frames etc are practical for installation in field). Any system would need to be an economical method (i.e., bidders could estimate the cost of implementation across hydrosystem, plus include any information that could affect project costs such as outage length, etc.). Lee estimates that a solicitation could cost around $840k plus lab costs (he uses ~$1M as a rough estimate). The results would yield greater confidence on how to proceed.

Tom Murphy requested that HDC calculate the value of a more accurate method compared to the cost of testing. There are a number of questions that need to be answered before proceeding with a solicitation, such as: what is the delta between absolute and relative flow measurement? How would dispatch change? Can TDA data be used? What additional benefits would be available if more accurate data is in T2? Will the cost of the solicitation yield sufficient benefits/results? Maybe this proposal could be considered as R&D or applied research. HDC could see if there is interest from other organizations.

There are two known systems of measurement, so it should be possible to calculate the costs of implementation and compare that to the value of benefits to be gained in T2 from increased accuracy. Tom requested that Lee document the economic benefits from increased accuracy from absolute flow measurement based on known operating conditions/constraints in T2 (using TDA unit 14 as a data source).

Follow-up actions:

· Lee Sheldon will add language to the “applied research” absolute flow measurement solicitation specification that insures that HDC considers and tests only systems that are affordable and distribute it to the team for review; include provisions for a minimum equipment apparatus as well as the standard apparatus. [Note: this material was sent to the HOT on 10/12/06.]

· Lee Sheldon will prepare an economic benefit study that supports the cost of the proposed applied research project in terms of the expected efficiency improvements that could be realized in the FCRPS in light of all known operational constraints.

Absolute flow Accusonics at LWG/DWR

Rod Wittinger led a discussion of the existing contract and subagreement scope and funding. Additional work on a data acquisition issue is still needed to incorporate the results into GDACS.

Machine adjustment/health check proposal

Ed Miska reported that there is no update.

Dynamic blade angle measurement

This project has been completed, and the proposal has been approved for funding.

3-D cams

Currently the electronic 3-D cam systems installed on the Lower Columbia (BON, TDA, JDA) differ from those installed at MCN and the LSN. The Lower Columbia cams are the older original stand-alone NWD cam and the NWW cams are a newer generation design. New cams use superior gate and blade position transducers, have a number of design improvements that allow cam tables to be easily changed, and are installed in the GDACS RTU. Now that the proposed plan is to interface the T1 test box with GDACS, the Lower Columbia cams will need to be changed to accommodate the planned design of the ITB and need updating anyway to be able to use T1.

At this time, additional information is not yet available for a proposal to be presented to the HOT. Richard Nelson stated that there is general agreement that new cam curves should be built into the new digital governors, but now the governor project is delayed. Should the time and costs (~$160k for NWW) be spent now to update the cam curves into GDACS, and then transfer them later into the new governors? It could pay off to spend the money now.

It was noted that there is an issue of GMC resources and funding, and this updating was not planned in GMC, so it should be supported separately.

Follow-up action:

· Robert van der Borg will develop estimates to update the Lower Columbia cams.

Action items (Exhibit 9)

The action items were reviewed. Tom Murphy will update the action items based on this meeting. Following are a few highlights:

Augmenting FTE to work on T1: HDC hopes to have FTE on board soon. Ken Earlywine will give the date to Robert van der Borg.

RCC and NRTO: the Fran Halpin, Tom Murphy, and Ed Miska will work with Don Faulkner who will work with projects.

Next meeting

The next meeting of the Corps Hydro Optimization Team will be held January 23, 2007 in Portland. The all day meeting will start at 8:00, and will include a report on how T2 performs against expectations and feedback from the “customers” (i.e., operators) with an anecdotal summary or survey. No date has been set yet for the next Reclamation HOT meeting.

· LIST OF FOLLOW-UP ACTIONS

Reclamation action items are shown on pages 1 and 2 of the text of the minutes.

Corps action items follow:

· HDC will confirm whether the GDACS tables need to be updated for T2. (flow table update)

· HDC will report whether T2 works in the field and whether the operators are happy with it.

· Tom Murphy will push TBL to develop to meet the ICCP deployment date (Tom-lead; Ed for design and configuration; with Kent Anderson, Steve Davis, Dick Hammer, Scott Bennett, and Glen Smith).

· Clarify path for approval of 3-D cam curve information and performance and flow information.

· Robert van der Borg will develop estimates to update the Lower Columbia cams.