1. Harold Kirkham Jet Propulsion Lab 21 May 2002 The NEPTUNE Project: Wiring the Juan de Fuca Plate for Science Bruce Howe University of Washington

2. NEPTUNE’s Origins The challenge: Move data to/from all over the plate 2 shore connections, but 30 nodes Get power all over the plate Have incredibly high availability Use only telecomm cable 30 year lifetime goal “Maintenance-free” system There is a need for long-term observation on the scale of a tectonic plate

3. This part of the talk: The plan: Give the background Describe the solutions chosen Discuss known problem areas Invite discussion with the audience Aims to tell you what engineering work is going on A good deal of NEPTUNE calls for original thinking YOU can help!

4. Science Objectives Constraints Communications (WHOI) Power (JPL) Data (HIA) Time Distribution (WHOI) Project Management (JPL) Architecture & Requirements NEPTUNE’s Origins

6. NEPTUNE compared to SPACECRAFT Deep Space –Long life –Environment –EMC –new and inherited –robustness/cost/ science/operations trade-offs –Failures costly and/or impossible to fix Deep Sea –Long life –Environment –EMC –new and inherited – robustness/cost/ science/operations trade-offs –Failures costly to fix often with significant delay

7. Node layout

8. Communications The differences: Move data to/from all over the plate 2 shore connections, but 30 nodes Include video originating under water Maintenance differences The solution GbE over single-mode fiber Routers instead of repeaters Redundancy in the nodes This is unlike ordinary submarine telecomms Capability of COTS systems: nodes ~ 100km apart

9. Communications

10. Sometimes the routers will fail Primary comms must have a “back-door” It was realized late that the back-door could also be a backup comm scheme The solution SAIL over dedicated fibers Back-up communications The Power System will operate with reduced performance but it is not clear that data can be handled at all

11. Power It not like existing sub-sea power systems It is not like existing land-based systems It is not like historical power systems either If power doesn’t work, nothing works! The NEPTUNE power system is novel:

14. Series or parallel? Series: difficult to branch into a network current same all over – efficiency poor Christmas tree light problem Parallel: easy to branch into a network current varies – efficiency high So of course, undersea cables are all operated as series-connected systems

15. Frequency 60 Hz: very familiar technology cable capacitance real problem 0.1 Hz: needs power electronics (just like dc) unfamiliar technology probably good for the cable dc: reasonably familiar technology needs power electronics hard to make a circuit breaker probably not good for the cable

16. Frequency 60 Hz: costs too much to compensate 0.1 Hz: unfamiliar technology (too scary) costs too much to do energy storage dc: wins by default probably means derating the cable

17. Voltage the higher the voltage: more efficient more power! scares the non-power guys shorter cable life? 10 kV wins because: nice round number it’s a common value in submarine cables

18. Converter principles

20. Converter trade-off the fewer the stages in series: the easier the MTBF requirement the lower the frequency the larger the components 50 stages wins because: nice round number for the voltages it allows a high frequency and small size However: MTBF number will be a challenge testing will be a challenge

27. converter redundancy

28. measurements there are many interesting measurements to make for the control and protection systems: current on the MV side (an isolation issue) voltage on the MV side (a dissipation issue) leakage parameters of interest include: accuracy stability bandwidth

29. measurements relevant current measurement technologies include: Faraday Effect Hall Effect optical power relevant voltage measurement technologies include: resistive divider all current measurement schemes

30. Measurements: Faraday Effect

31. current injection voltage output magnetic field Measurements: Hall Effect Lorentz force pushes carriers aside

32. Measurements: Optical Power

33. Measurements : Comparison of methods Faraday EffectHall EffectOptical power expensive cheap not cheap highly accurate not accurate highly accurate very wide band good bandwidth very complex very simple fairly complex Hall Effect is the winner! needs no Fe needs Fe circuit uses series R

34. Measurements: leakage detection need to detect leakage current on the load side one option is to isolate the output, and ground it through a resistor another option is to measure the current difference between the two wires isolated source

35. Measurements: Current Comparator