Data Driven O&M Strategies with Wattch and Omnidian | EP232 - Clean Power Hour

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Alex Nussey (00:00):
If you're going to make a truck roll making sure
that it is maximally valuable,the worst case is diagnostic
truck rolls. If you have to rolla truck out to site to figure
out what the problem is, thenyou're going back to fix it
later. You probably don't havethe right replacement part. It's
about making sure that the rightpeople come to site. Depends on
what needs to be done. So by thetime you roll a truck to site,
you should know that the issuecannot be resolved remotely.

(00:22):
We'll touch on that as well, andthat you're bringing the right
person in the right part thefirst time. And that's that's
super that's super, superimportant.

intro (00:31):
Are you speeding the energy transition here at the
Clean Power Hour, our host, TimMontague and John Weaver, bring
you the best in solar, batteriesand clean technologies every
week. Want to go deeper intodecarbonization. We do too.
We're here to help youunderstand and command the
commercial, residential andutility, solar, wind and storage

(00:51):
industries. So let's get to ittogether. We can speed the
energy transition.

Tim Montague (00:59):
Welcome to supercharging solar O and M with
data driven strategies for rapidgrowth. I'm Tim Montague, your
host. I'm just thrilled to haveAlex Nussey With Wattch and
Trish Graf with Omnidian with ustoday as our expert panelists,
and we're going to do a deepdive into how O and M is

(01:21):
changing because of the adventof modern technologies, like the
platform that Wattch hasdeveloped. So I look forward to
learning more about thatplatform, and you can type your
questions into the Q and A orthe chat. I think the Q and A is
preferred, if that works, dothat, and then we will answer

(01:45):
those, either on the fly or atthe end, at, say, quarter of but
we will get to your questions.
And we really want to hear yourquestions. That's how you're
going to get the most out ofthis webinar, is by asking us
some questions with that, I willlet Alex nussi introduce himself
for the audience, please.

Yeah, absolutely.
Thank you, Tim. Super excited tobe here, and thank you, Trish
for joining. Really, really beenlooking forward to this for a
while. My name is Alex Nussey.
Mostly go by Nussey For those ofyou that have met me before, and
I'm the co founder and CEO ofWattch. Wattch is a monitoring,
intelligence and controlplatform for renewable energy,

(02:29):
so solar, energy, storage, Ev,charging primarily, and we are
we work very closely with O andM providers, both EPCs and
dedicated owning providers tosort of help maintain the
maximum performance on theseassets. I guess that's my that's
my background to doing this forabout five years now, as much as

(02:51):
I do love listening to myselftalk, I'm going to reiterate
what Tim said, please askquestions. Really want to make
sure that we're sort of engagingwith the audience. I think these
things are always more fun whenit's sort of a dialog. So thank
you.

Thank you. Alex and Trish Graff,

Trish Graff (03:04):
hi Trish Graff, VP sales and strategic
partnerships, and I'm nydian.
Nydian is a technology forward,data driven operations and
maintenance services companythat provides performance
assurance for our clients andpartners on their PB assets and
also their their battery storageassets. So thrilled to be here

(03:25):
as well, and likewise, pleasebring the questions forward.

Are your solar and storage assets performing at
their peak? Some wait forfailures or send technicians for
costly diagnostics. But what ifoptimization was just a click
away. Enter wattch, the cuttingedge monitoring and controls
platform with wattch owners andoperators can maximize their
portfolio performance, avoidunnecessary site visits, boost

(03:51):
production by up to 8% allwithout expanding your team.
Ready to supercharge your cleanenergy assets? Visit W, A, T, T,
C, h.io, and power up yourefficiency today. Fantastic.
Well, I'd love to give ouraudience a setting of the table.
What is the state of the stateof monitoring and analytics when

(04:16):
it comes to distributedgeneration with with an emphasis
on CNI and small utility solar.
So Alex, why don't you set thetable a little bit and then I
also want to hear amidians TrishGraf's comments on where is the
industry and what are some ofthe challenges that the industry

(04:38):
is facing.

That's a good question. I think that there's
been so much focus for so longin the industry on the sort of
leading edge of the growthcurve. We're trying to get new
assets on roofs and in theground. And when I joined the
industry, at least, it seemedlike there wasn't as much
attention on. In O and M, and atleast especially not efficiency
in O and M. But as theseportfolios continue to grow, and

(05:05):
importantly, start to age, it'sreally becoming a lot more
important. And the kinds ofproblems that you could solve
with sort of just man hours witha portfolio of five or 10
projects, it just doesn't holdup when you have a portfolio of
hundreds of projects. I thinkyou mentioned sort of CNI and
small utility we would say likeDG scale, distributed generation

(05:26):
scale. I think that's certainlywhere wattch sees ourselves
participating most heavily inthe market. We do serve some
utility scale customers, but thesort of core of our customers
portfolios is in that large,commercial and industrial to
small utility range. And I thinkthat these are, I think the
right way to put it is any sitethat is not large enough to

(05:47):
warrant full time staff on site.
If you're building a halfgigawatt project, you probably
have somebody there every hourthat the sun is up. But that's
not true with a distributedcommercial portfolio, and that's
where sort of leverage of Trishuse, you know, the term data
driven and technology enableddecision making. I think it's
where it becomes. It's not justsort of the nice to have, it's

(06:08):
required to do a good job atmaintaining the sort of the high
bar these assets. So I think ourgoal is to help the industry
continue to scale withoutneeding to scale headcount as
quickly to maintain these, thesegrowing portfolios, I could talk
on that thread forever. And

just for some context, we installed, I think,
something like 35 gigawatts ofsolar in the US in 2023 and and
Trish. What is the scale of ofyour business at Omnidian?
You're one of the major O and Mproviders across the country in
the DG space. But tell us alittle bit about your portfolio
and how you see the DG industrygrowing.

Yeah, and perfect dovetail off all the pieces that
Alex brought forward is we'reseeing a growth in DG for a
couple of reasons to bring backto your earlier point, Tim is,
is, you know, the IPPS, EPCsowner operators are looking to
grow their megawatts. And largescale. Utility scale projects

(07:12):
have their own set of challengesand issues can take a longer
time. So the shift to deploy fora utility scale owner, more DG
has really been driven by thecycle can be quicker, right? We
can get megawatts up faster withthat. As Alex said, the
challenge is, well, how do youresource? How 100 a 500 megawatt

(07:35):
solar, utility scale solar,looks very different when you
have 500 megawatts of DG. Theseprojects are scattered
nationwide. They're not largeenough to warrant a full time
resource at that asset. And sowhat we've done in in our offer
and our focus is to support thefast growth scalability of

(08:00):
large, scattered, disperse DGportfolios, so that anyone
that's coming into the marketcan resource up quickly. You've
got an asset in Hawaii or one inPuerto Rico with on Midian, but
doesn't matter, we can supportyou immediately. So, and we're
seeing that too, like withacquisitions, right? There's a

(08:21):
lot of buying and sellinghappening in in our in our
space, and for an owneroperator, IPP, that's looking to
acquire projects to help growtheir portfolio, our goal then
is to make that really simplefor them, so they don't have to
add headcount to manage allthose projects.

Yeah, so we've got this context of rapid growth and
limited resources, human power,time, money, and we're trying to
keep these assets performingoptimally. We're trying to have
an efficient machine to monitorand analyze the data coming out

(09:05):
of the field. When do we need toroll a truck? When do we need to
reboot an inverter? And it'sjust a big, hairy problem, and
it's growing very fast, so whenyou guys think about these

(09:25):
challenges that we've kind ofput our finger on, and you, Alex
and your and your company aredeveloping this, this kind of
next generation platform formonitoring and analytics that
will lead to better O and M,more productive operations. Tell

(09:47):
us a little bit about how yousee that landscape and and where
are you really leaning in to.
You know, this, this piece of adata driven O and M, I.

So I think that there's two sort of steps to
maintaining this, this diverseportfolio like this, and the
first is driven by the fact thatnot only are all of these assets
sort of geographicallyphysically distributed, they're
also very heterogeneous. None ofthem are the same. They all, you
know, especially when you havesort of a network of VPCs

(10:21):
building for one owner, you'regoing to see different modules,
different inverters toeverything at every one of these
projects. And as they age. Forthese smaller projects, you
probably don't have a strategicreserve of replacements. You're
probably buying new parts thatdon't actually match. So now you
have multiple brands at one sitefive years in. So these every
project is unique from everyother project in the portfolio,

(10:41):
and they're scattered around. Sothe first step is getting a sort
of single pane of glass, abird's eye view over the whole
portfolio that makes it easy todraw conclusions and analyze
information very easily. Andthat's sort of what we see as
our first responsibility is getdata from the field into the
operator's hands in a way thatis completely standardized

(11:04):
signal. We call it our universaldata model, but transform all
the information as it's comingin, so that you can pivot on it
as easily as possible. Thesecond piece of that is finding
sort of signal in the noise wecapture for a good size
commercial, like a largecommercial industrial project,
will capture something like abillion data points in a year

(11:26):
from that from that site, and nohuman is going to go through and
look at all of that data, right?
So it's, it's that sort of wherethe machines step in to crunch
that down and help you take thefirst step, tell you where to
look with the end goal ofhelping you identify as closely
as possible precisely whatchanges could be made to improve
performance, or what hashappened to cause a reduction in

(11:48):
performance. Sort of, from thereit comes the, you know, famously
hard problem of the travelingsalesman. How do you actually
get to all these sites? Whogoes? When do they go? And
that's, that's not somethingthat we winner is involved in.
But in terms of, sort of, Ithink the intelligence piece is
all about making it as easy aspossible for operators to

(12:09):
identify and size issues. Isthis a $50 a day problem, or is
this a $5,000 a day problem? Youknow, it's sort of localizing
that and then making sure youknow what you need to bring with
you when you go out. Do you needyour replacement part? Do you
need this kind of personnel onwhat's, what's the issue before
you roll the truck? I'll

stop there for a second, yeah, and Trish, you
know, I'm curious what Omnidianthinks about how assets are
performing overall, and thenwhat is the the value add, right
when you onboard a moresophisticated platform like
Wattch, yeah.

So the first part of that is, you know, our job as
the custodian. I mean, we wetake care of our clients assets
as if they're our own right. Weare. We measure ourselves on
performance. So, as Alex said,like the components to really
identifying that is the datathat's coming out and the
analytics that sit on top tohelp really sort through all the

(13:11):
components and the noise. Isthis, it really a problem that
needs to be addressed. Is thisreally? Is this affecting
performance? Or is thissomething in when you're
starting a project, you need totake into account for your true,
expected energy, right? Where inthe US is this project located?
You need to account for weatherand things like that. And then

(13:34):
the second piece is for us,it's, you know, taking
partnering and having a dasprovider, a data acquisition
system such as Wattch, reallymakes our jobs easier. We act as
an extension to our client'steam, right so we're technical
asset performance managers, andhaving the right data with the

(13:56):
right frequency, with the rightcomponents built in really
allows our platform, then wehave our own proprietary
platform that sits on top of theWattch data to further enhance
the components of what is theactionable item that needs to

(14:17):
happen here? Does it need tohappen? Does it justify, as Alex
said, like if this is a $50problem, but the cost to send a
truck is going to be 1000 themath doesn't work. And so we
take the information from from aprovider such as watch, which
has an amazing platform. Also wehave performance engineers that

(14:40):
look at it to help determine howmuch of the performance is this
affecting, to really consultwith the client on what is the
best choice for this asset, andyou know your OPEX or the
portfolio, I think

we like the word bionics here at Wattch. I think
it's. Not, it's not totallymachines. It's creating tools
that help the right people whoare who are actually doing the
job coding leverage. But bylikes is one of our like Wattch
words here. So

I remember in our in our pre show discussion, that
I think of this in terms ofdashboards and humans. And, you
know, in a perfect world, thedashboard is greatly
facilitating how humans thensort through the information

(15:31):
coming in from the field andprioritize the information, and
as much as possible, it'sautomating that for me and
saying, Okay, here's the greens,here's the yellows, and here's
the reds. The reds are needingimmediate attention, and that's
your priority, human in the inthe DAS center, or the knock

(15:54):
right? And then they'redeploying assets, whether that's
remotely or boots on the groundto fix the problem and get those
reds to be greens. How do youguys see that? And how does
Wattch facilitate thatprogression from bits and bytes

(16:16):
coming from the field to acomputer screen to a human army,
human army. I like,

like that. I think the sort of the separation where
we at least draw our line ofdemarcation is we're really
looking at the technicalperformance element. So if you
have, you know, a $500 problemand two $200 problems, you know,
or two $300 problems, but youcan hit them both in one truck
roll, you know, maybe itactually makes sense to do that
one instead, right? It's, it'sit's when you, when you identify

(16:45):
and size the problems, and I'lltalk more about that in a
second. But that's, that's sortof where we start, and then it
comes into the real world ofhumans and where they are today
and and what they're doing.
That's where, sort of managingthe army is something that we,
we work with fantastic folkslike Omnidan. For us, we think
it's getting that green, yellow,red, is our primary

(17:07):
responsibility, and we kind ofstart with that. Trish used the
word expected production, orexpected energy. So a common
term in the industry, that'swhere we've spent a ton of our
engineering resources on, is ondeveloping this model. We call
it digital twin, but you cancall it whatever you'd like. To
tell you what a system should bedoing at any at any moment. And
while there's a top line there,you know, on a monthly basis,

(17:30):
what should I be getting?
There's also the drill down,which, in my opinion, is where
the real utility starts to comefrom. Is it's not just at the
site level, not just at theinverter level, but like getting
down to every string on everymvpt of each inverter, you know,
oh, is an RSD failed? Like, canwe detect that? Because the
voltage in the string has gonedown, you know, is is this

(17:51):
misconnected? Is there soiling,reducing the current? And we
have a set of workflow toolswithin the product that help you
sort of say, Hey, look at thetop line. Your site's only
producing 95% of what it shouldbe. All of that is attributable
to one of these inverters. Andit's actually an issue on
strings two and three. And thenit's like, okay, go, point the
right direction and tell youstart here. That's sort of why

(18:13):
that is, sort of humans come in.
Was it a mistake that didn't getcaught at commissioning? Was
there a hailstorm recently? Andthat's there's a fine line with
any dashboarding tool betweenAutomated Insights and then
still giving the operator accessto the raw data. You really want
those to live side by side, sothat the operators know where to
start, but then can jump in andlook at as high fidelity

(18:34):
information as they want to sortof help draw those conclusions.
I don't know if that actuallyanswered your question,

but Trish, you're you know, Omnidian is inheriting
portfolios at various stages ofof their lifespan, from newly
energized to have been aroundthe block operating for some
years, and then you're chargedwith improving performance, and

(19:03):
you're actually guaranteeingperformance at some level to
your customers. But tell us alittle bit. You've got some
wonderful portfolios that aregreat examples of how big and
hairy this problem is, you know,portfolios of 100 plus megawatts

(19:25):
of DG projects that are spreadover multiple states. And I
don't, I don't, I don't cherishthat responsibility. It's, it's
a massive responsibility thatmeans real, you know, real
revenue potentially lost foryour customers. Tell us a little
bit about how omnidian tacklesthis, and why did you, you know,

(19:51):
onboard Wattch, and what is theend result that a customer can
expect when they work withomnidian. I. Um, with the
benefit of Wattch underneath it,yep.

So I'll take that in kind of chunks. You're
exactly right, taking on largeportfolios of as as Alex said,
no two are alike. As far as youknow, the components and the
system type and size scatteredall over different ages,
different custodians, to Tim. Soyou may have a portion of the

(20:23):
portfolio that was under greatcare and great Wattch, pun
intended, and some that werenot, and I think which starts
with when we're aligning withour partners, our clients around
you know we're our job is to,yes, ensure performance and
increase performance. The firstpiece is aligning at, well,

(20:45):
what? What do you expect thissystem to do now? And as Alex
said, we're watching Omnidianare very closely aligned on
building a digital twin. Weabsolutely need to make sure we
see your system as it's trulybeen designed and built. So it
starts with, there'sopportunities to Tim still

(21:06):
definitely in the marketplacefor these older projects, to
really kind of calibrate on whatshould it be doing? Good news is
we get there. We have ourperformance verification tests
that we do systems online for 90days. Sometimes that can create
a lot of heartburn with theirclients, because they'll have

(21:28):
had a number in mind about whatthis should be, should be doing,
and 90 to 120, days later, we'regetting that actual real data on
this is what it is. So worstcase scenario, our calibration
point happens after that,performance verification,
testing, and then ourcommitment, as you mentioned,

(21:49):
also with our performanceguarantee, is that we've aligned
on a number, and we'll offer a95% non weather adjusted
performance guarantee. No oneelse does it, very different
than an availability guarantee.
So we're focused that thatinherently drives our activity
set, and our focus on ensuringperformance right system can be

(22:11):
available at 10 o'clock atnight. It doesn't matter. It
can't produce anything. So allof these, these components, lead
to setting up Omnidian As we arefully driven, and we've got skin
in the game now with theperformance guarantee that we're
going to work with you to ensurewe're taking the right actions

(22:33):
for the right the right actproblems that are really
affecting performance. To goback to what Alex said earlier,
there can be faults and alarmsand and if you have somebody
internally on the client sidethat's watching these they're
probably not trained like aperformance engineer would be to

(22:54):
be to to understand what is thisreally affecting. We see a lot
that without that true, theright data with the right
analytics. And then, as yousaid, Tim, the human who's then
assessing all of that todetermine, does this justify
going out to make, to make achange of repair, a fix you

(23:16):
could be sending trucks everytime a beat goes off and you are
upside down pretty quickly. Andso for that with Wattch, what
has been great for ourpartnerships with our shared
clients, is that we know, from adata acquisition standpoint, how
Wattch is, how they're grabbingthe data, the different time

(23:38):
stamps that they're grabbing itfrom, which components they're
looking at. And as Alex said,string level information, all of
that. Those details help us dobetter on the performance end,
on making the right decisionsfor the asset.

So there's, there's three big buckets that
we're talking about here,efficient resource allocation
and prioritization, remote issuedetection and diagnosis and
minimizing false alarms andmissed problems. Alex, tell us
about those buckets, and if youcan give a real world example,

(24:22):
you know that would be, thatwould be amazing. But how does,
how does watch see this world,and how does it come into play
in the real world?

Yeah, yeah, that's definitely can talk about that.
I think the three buckets youmentioned, I will let Trish
mostly speak to number one,efficient resource allocation
and prioritization. And I'llactually, I want to zoom out for
a second. We didn't mention thisup at the front, but I think
that bucket to their remoteissue detection and diagnosis is

(24:54):
something where Omnidian andWattch overlap. Omnidian
obviously works with a lot ofdas providers. Wattch obviously
works with a lot of on. Folks, Ithink we both really enjoy
working with each other, becausewe have this sort of shared
ethos that makes it very highleverage when we when we have
these shared clients,particularly high leverage so

(25:15):
that that second bucket, remoteissue detection and diagnostics
is is something that we sort ofhave, especially to back after
that the diagnosis we have somesome overlap in I will touch on
a couple of real world exampleshere. I think we see issues come
up in sites two times. They'resort of in the ongoing
operation, which is whateveryone thinks about, the new

(25:37):
problems that emerge. And thenthere's commissioning, making
sure that Trish used the phraseas built, making sure that
you're as built is actually whatwas as built, especially in
commercial getting aauthoritative representation of
what was installed on site andhow it was installed can
sometimes be a painful process,but it's essential in order to

(25:58):
right size expectations with thelong term owner. So I'll start a
little bit couple of real worldexamples. And if you don't mind
sharing that those slides I sentyou, I'd show a couple of
different items from that.

Sure. Slide One is on screen now, cool, yeah, so if
you don't

mind jumping to the third slide there, okay, it's my
little face, actually, yeah. Sothis is just one of the ways as
an example of a sort ofscorecard that Wattch provides
for an inverter. So we look atits at the very top level. We
have the overall health score,as we call it, that actual
versus expected, or actualversus ideal energy. So this

(26:37):
inverter is very healthy, and wesort of analyze an inverter in
three stages, mostly by lookingat the DC inputs from that
photovoltaic array. So the firstis, and I actually realize you
can't see my mouse, but I'mlooking at this first table
here, the sort of ideal versusmeasured voltage. And this is a

(26:57):
great example of sort ofautomating steps for human users
to have help give them the mostleverage possible. So that
measured voltage column is anaverage of the input voltage on
each of the strings for thatinverter over the course of a
day, and then that ideal voltageis what our digital twin says

(27:21):
the voltage should be. Thistakes into account a number of
calculations. We're looking atthe irradiance at the
pyranometer. We're transposingthat onto the array. We use the
IV curve of the module and itspan file, and we essentially
simulate what its voltage shouldbe. And then when we compare
those, we can see that they veryroughly match. Now if you go

(27:44):
down one slide, you'll see,here's an example of another
inverter where they don't match.
In this case, MPPT is four andfive. Looks like we basically
swapped those two strings. Andin this case, that actually has
a pretty minimal you can see 96%health score in the top minimal
impact. But depending on howthose are switched and how many

(28:05):
you've put on, each string, canhave a material impact on the
overall site performance. Andthis is, this is something that
within, you know, 10 or 20minutes of turning the site on,
you can come look at this andmake sure that you have strung
the project correctly. This issomething that we really
encourage. Oh, really encourageowners to use as part of
acceptance when they're sort oftaking over the project. And

(28:26):
this is just one of those kindof analytical workflows that I
mentioned. We take the raw datain and that's that measured
voltage, we augment it withsimulations from our digital
twin, and we give a sort ofrough sized conclusion to the to
the operator using the product.
We have that final column therelength mismatch is basically how

(28:49):
far off we think the strings areand from being wired correctly.
Now the owner can then ask tomake a decision on whether or
not it is worth fixing, andthat's sort of the next step,
but it is our goal to put theseconclusions within five minutes.
Don't need a lot of data to runthese analytics. We can sort of

(29:10):
in a very automated fashion,tell you whether what got
installed actually matches theas built drawings that you
received, and from here, you fixenough small problems, even at
one or 2% so invertersperforming at 96% it adds up and
across the board, on average,you you can get some pretty

(29:31):
material gains. We see for ourcustomer base, for commercial
and DG scales about 8% gain forpeople that start using these,
these tools to commission, andthen, in an ongoing fashion,
maintain their, maintain theirsites. Um, let

me, let me ask you a couple of questions here. So
when, if you're, if you'relooking at a site, you're,
you're, you're sitting in aknock and you've got some flags
on sites, and then you're, you.
Drilling into the site andlooking at a more granular
level, you I immediately amconcerned that the number of of

(30:07):
assets involved is is verylarge, and so is, is Wattch
though filtering through thisinformation and helping the
human at the knock level knowwhere to drill down. Or is it
simply just flagging sites, andthen it's up to the human to

(30:29):
determine priorities? Yeah,

so and I'll actually dovetail. I think I can
see somebody put in the chathere. I was looking at the
questions that I see in thechat, faults versus alerts. Um,
so I think we, we think there'skind of two stages to
performance management. The onethat's hardest is when you're
bringing the sites on for thefirst time. It's that like right
sizing of expectations. I'lltalk about that in a second. In

(30:54):
the steady state you shouldreally be push more than pull
the product. Our product willtell you when there's an issue.
We if you have everythingperforming well, then we're
constantly watching every 10seconds as data flows in from
each site. We're looking at thatdata and reassessing whether or
not it's still healthy, and ifit is not, we can create an

(31:14):
alert, a notification, for theuser, and that says, hey, maybe
go take a look at this in termsof prioritization, if we have
the financial implication ofperformance at that site set up,
we can actually rank thosebasically by estimated daily or
monthly cost. So how much is howmuch is this in dollars, going
to impact the customer, whetheror not you make a resolution

(31:38):
step comes down to how much it'sgoing to cost you, and that's
where, where it is, where mytrucks are, where my boots are.
And that's sort of outside ofthe scope of our product, but
the cost that it is causing issomething. So we go from healthy
asset enters an unhealthy state,create an alert, so that the
user sends an email, sends atext message, whatever it is, so
that the user knows to to gocheck on it. And that's when we

(32:01):
enter one of these workflowswhere it's like, okay, there's
underperformance. We know howmuch it costs. Let's narrow in
on why and But that's after sortof we've already created this
notification and and roughprioritization for the for the
user. The more painful side iswhen you onboard a new
portfolio, you bring in 50sites, 100 sites, and lot of

(32:21):
time in Wattch. When we do thatwith especially in CNI, it's all
red, it's all F's and D's, andit's that conversation of, you
know, some of these are olderassets. They have a lot of sub
critical issues that you know,anyone can catch when an
inverter fails. But if you havean nppt that's coming in and
out. Most inverters, mostmonitoring products, won't flag

(32:43):
that kind of problem. But all ofthat adds up, and across the
board, you can have a prettyunhealthy portfolio. And then
it's the conversation of, youknow, these are 10 year old
assets. Does it make sense to gospend the money to bring these?
They're at 70% you know, itmight be two thirds of the you
know, you might be able to getto 95 in a third of the cost,
and it's another two thirds toget to 100 is that worth it? How

(33:04):
do I want to set my expectation?
And that's, that's the most timeconsuming part, I think, in and
setting, you know, expectationsetting little sensitive.

The Clean Power Hour is brought to you by CPS
America, the maker of NorthAmerica's number one three phase
string inverter with over sixgig Watts shipped in the US. The
CPS America product lineupincludes three phase string
inverters ranging from 25 to 275kW their flagship inverter, the
CPS 252 75 is designed to workwith solar plants ranging from

(33:39):
two megawatts to two gigawatts.
The 252 275 pairs. Well, withCPS America's exceptional data
communication controls andenergy storage solutions, go to
chintpowersystems.com to findout more. Heller asks a good
question, what part dodiagnostics, faults and alerts
play in the O and M datacollection, and which are the
best. And I guess I'm curioushow Trish, how Omnidian, thinks

(34:02):
about these things. Becauseyou're, you're really on the
receiving end of this.

Yeah. So it's also too, it's, it's the
nomenclature, right? So it's,it's, to some of our clients, an
alert and a fault can be, likethe same thing and that and lies
where we step in and Wattchsteps in is depending on how you
you have your your gas set up.
You, you could get 1000 alerts aday, and no, no one human is has

(34:30):
correct, and a lot of it is, Idon't want to say noise, but
most often noise. And so it'sthe ability to have the
analytics, or the AI and the MLthat sit in there to really sift
through, and as Alex said,having that front end financial
information to also really pursethrough what pieces are matter

(34:51):
to the to the ROI right, andanalyzing the root cause of the
issues and. Having the rightdata and the right analytics and
the right performance people andteam to look at it is really, to
me, the the cherry on top, so tospeak, so that you can
determine, all right, we'regoing to filter out these

(35:13):
temporary conditions. Shading isa good is a good example? Oh,
it's underperforming, is it? Oh,it went back up. It went down.
It went up. Why? Well, the one,there's nothing to do about
shading, other than if it'sdepending on why, right? Is it
another building coming intoplay or something, or are they

(35:34):
alerts that don't requireintervention? So these systems
can be very finicky and verynoisy. And what the Wattch
platform does, and then what theOmnidium platform does is,
again, is get rid of the thingsthat don't require to send
someone out. What can we doremotely to potentially? That's

(35:55):
always our first step is, whatcan we do remotely to address
the situation. Our goal alwaysis to reduce, minimize,
eliminate the OPEX costs of ofsending a truck out there. And
we've, we've, we've proven iton, you know, 500 megawatt
portfolio. Tim, an interestingstatistic is that the industry

(36:16):
benchmark for truck rolls permegawatt is, is six truck rolls
per megawatt, your numbers, youryour business case is, is going
to be upside down prettyquickly, over

what, over what period of time when you say six
truck, six truck rolls permegawatt,

like, think the data that We used was 12 to 48
months, okay? But we then eventruncated it to let's look at
just 12 months. If the industrybenchmark is for the five year
life of an asset is six truckrolls per megawatt. What? What
is Omnidians participation inreducing that? And let's even

(36:55):
just say, the first year, it'sunder our our custodianship, and
we were able to reduce that byover two thirds. And that's
because, well, this is ourwheelhouse, right? So, so with
the right data and the rightanalytics and the right
performance team looking at it,it is, don't send a truck every

(37:16):
time. Send it only when it'sgoing to it when it's necessary
to really make, as Alex said, animpact on the financials. And I
think we're going to get therelater on the conversation. But
Alex knows this. Tim, youprobably know, I'm sure you know
this as well. That number has tocome down. It can't be six truck

(37:38):
rolls a megawatt. There's no waythe financials on these assets
are going to work out long term.
And there's a labor shortage. Wehaven't even touched on that
yet. From a technicianstandpoint, there's a labor
shortage on the qualifiedexpertise that is required to go
and do these repairs. Like Isaid, I said, I think we'll get

(37:59):
there a little bit later, butthere's lots you can do early on
in a project to avoid all ofthat activity and going to the
site. Yeah,

I guess it really boils down to two things,
improving performance andreducing the cost of O and M.
Those two levers are going topush asset ownership in the
right direction, right? This isgood for the bottom line of the
asset owner. So let's, let'sdrill down though a little

(38:32):
little bit more about let'sassume, okay, for a second, that
you have good construction thatthe project got off on the right
foot. That's not always thecase. Certainly I could live in
that world.

Unknown (38:47):
Yes,

and maybe you can comment on, you know, what
percent of projects get off onthe right foot, but then there's
going to be degradation overtime. Things happen, connections
come loose, equipment breaks,shading happens. Maybe there's
not proper vegetationmanagement. Maybe there's a dust

(39:14):
storm and there needs to be eyeson getting a cleaning done. But
what does Wattch allow Omnidianto do that moves the ball down
the field. I

think, I think it's about, uh, how should I say, if
you're going to make a truckroll, making sure that it is
maximally valuable, if you theworst case is diagnostic truck
rolls, if you have to roll atruck out the site to figure out
what the problem is then you'regoing back to fix it later, you
probably don't have the rightreplacement part. It's about
making sure that the rightpeople come to site depends on

(39:48):
what needs to be done. So by thetime you roll a truck to site,
you should know that the issuecannot be resolved remotely.
I'll touch on that as well, andthat you're bringing the right
person and the right part thefirst. Time, and that's, that's
super that's super, superimportant, and in terms of
remote capabilities, I thinkthat is getting better Wattch.

(40:08):
We call ourselves a hybrid SCADAplatform over a das platform,
because we have our like taglineis, you know, bringing the
sophistication of utility scaleSCADA to an ease of use and
price point that works forcommercial and DG scale solar.
So we want to bring as many ofthose remote control
capabilities that exist in thehigh end projects down to these

(40:29):
lower end projects as well. Sowhenever possible, if you can
complete a task remotely, let'sdo that before you go out to
site. Whether that's clearing afault, resetting an inverter, re
closing a main breaker,reconfiguring an inverter, if
you know you you install the newone and forgot to change a
setting. There's a lot that canbe done remotely these days. IV

(40:52):
testing is a new one. We nowhave. It's like two of the major
brands support remote IVtesting, and we support that in
Wattch. So instead of sendingsomebody out for a day to IV
test each string, let's justclick and grab that data from
every inverter while I'm sittingin my nice, air conditioned
office. And if I figure out thatsome of my RSDs are broken or my
combiner boxes, we can go aheadand get that new that new part

(41:15):
ordered before you make thetruck roll. So I think bringing
it back together step one, doeverything we can remotely full
control over the project. Wattchenables control for all the
assets that support it, and thatlist is improving. And then the
second is authoritativelyidentifying the problem, so that
you don't have to make nuisancetruck rolls. You don't have to

(41:35):
make diagnostic truck rolls. Youcan just go out and actually fix
the problem, walk directly tothe right place on site, and,
you know, make the change. ButTim, did that answer? Your
answer your question?

Yeah, that's good.
So we have a question from theaudience, what is the market
standard for performanceguarantee, and do you see
market? See the market catchingup to an Omnidians 95% can it go
any higher?

Oh, great question.
And I'll, I want to put a pieceout there, just for some put
some context is, is I have autility scale wind background,
and so what I see are lessonslearned, same thing in wind,
where these, these projects,lots of different parts, no two,
you know, lots of different OEMsand different standards, and how
are we taking care of them? Whatdata is important? So lessons

(42:24):
learned, even though completelydifferent technology, much
different scale and size. I wantto preface kind of how I'm going
to put put forward these nextcomponents with that. So we saw
in that older industry, whichstill not old, still growing. It
started with availabilityguarantees, right? I will
guarantee we see that in solaras well. I will guarantee your

(42:48):
asset is available. 95% soundsreally compelling, right? Except
for it doesn't matter much inwind, if that asset is available
when it's low wind season. Samewith solar. It doesn't matter if
that asset's available when it's10pm 9pm 8pm at night. So there
we've seen a shift with thereare performance guarantees.

(43:13):
We're not the only one that doesa performance guarantee. We are
the only one that takes theweather equation out, and I'll
say that, I'll package itshortly, is we assume the
weather risk. We have over 2billion data points. You know,
think of it as an insuranceactuarial. We are had we're
managing our risk on how isweather going to impact the

(43:36):
performance of that system? Wehave enough data to support
well, we'll take that on. Sothere might be another provider
out there confirming they'll doa 95% performance guarantee
challenges. What are the carveouts? Right? What's excluded?
Not just weather. What isexcluded? To you know what we we

(43:58):
really are trying to change theindustry to focus on the right
behaviors to get these assetsperforming. And so we say we'll
take on that weather risk,soiling, shading snow, like
we're accounting for all ofthat. We're going to confirm

(44:19):
you're going to get 95% nonweather adjusted, and we'll work
towards that together. If not,we we have a financial
reimbursement. We have to payback per kilowatt. How much did
you Did we miss? We owe youmoney. To circle that back, I
think the industry is, I thinkis still maturing, as far as

(44:40):
guarantees go. But again, if Itake my lessons learned from
from wind, it's it's justshifting and it's moving and it
Tim it all starts with, if youwant to guarantee performance,
you have to work up the chain.
The. Right parts, the rightconstruction, the right as

(45:05):
builts, the right digital twin,like everything has to start
upstream, and that's ultimatelyour goal, is to focus on, how do
we participate in making thisreally a renewable energy source
that can be really financiallyattractive.

So I love this question, what, what kinds of
things can you actually see andfix remotely? And one of the
contexts here is, I think it wasTrish's comment that, you know,
there aren't enough techniciansin the world to service our now

(45:44):
growing portfolio of solarassets. So this remote
monitoring and diagnosis isextremely important, but what
kinds of problems can weeffectively identify remotely?
And either one or both of youcan answer that?

Yeah, I'll take the first pass. I tried to dovetail
that question a little bit in mylast answer, but I think just to
rattle off, it changes a bit bycomponent. For inverters,
they're getting better, but justresetting them still clears a
lot of issues with with modernstring inverters like that.
Shouldn't have to happen, but itdoes so. Just being able to

(46:22):
reset an inverter can clear alot of a lot of issues. So
depending on your sort of,generally in the industry, we
hear das to mean renal readonly, and SCADA to mean sort of
read write to the site. If youhave that SCADA capability, you
can reset inverters. You can,you say like main breakers and
reclosers, most sites at thatsize are going to have SCADA.

(46:46):
Might be very expensive, but youcan reset that depending on the
allowed operating parametersfrom the utility. Same goes for
arc faults on inverters, resetthose, and then where things get
really interesting is runningsort of mutative diagnostic. So
an IV curve test is a greatexample. You can't gain that
information from an inverterwhile it's operating normally.

(47:08):
You have to put it in a specialoperating mode, have it sweep
over the input voltage range andcapture that data. And that's a
task that normally would need tobe completed on site. You de
energize the inverter, plug in apiece of test equipment, but if
you can control that inverterremotely. You can just coach it
through running that same test,all remotely, and that helps you
have one additional piece ofinformation before you go out to
site. So the answer is, youknow, depending on the if

(47:32):
something fails, it can't beresolved remotely, obviously,
but any problem that can beresolved without a physical
action on site. We let's, let'sdo that. And then if we do have
to go in and change somethingphysically on site, then let's
know exactly what it is by, youknow, introspecting, tweaking as

(47:54):
much as possible before you goout there. Another good example
is like waveform captures, ifyou have a power quality meter,
a lot of these revenue grademeters these days are actually
very smart. You can poke at themand say, Hey, capture me a
waveform so I can see exactlywhat's going on in sight. Pull
that remotely one less step todo in person. So I think, I
think that's the sort of thehigh end is solve some problems

(48:15):
and diagnose almost all others.
Really want to know what's wrongbefore you out there. Trish, do
you have anything you want toadd on that? No,

it's kind of what I'll put forward is really a
culmination of the conversationwe've already had. Is it starts
upstream. First of all, like,are you getting? We could talk
about, we remote diagnose 75% ofthe problems. Well, Tim, and the
reason is, is because we'vegotten they're not really
problems. You have to siftthrough that first layer first,

(48:44):
right? So that's the opportunitywhere a Wattch comes in and
partner with dominion, with ourAI and ML is to those six truck
rolls per megawatt are becausethe intelligence, the data, the
analytics, the performanceengineer isn't those eyes aren't
on that asset. You're set. Everytime you see something, you're

(49:06):
sending it. As Alex said,components are continuously
improving. Things you know, aregetting better. We can see more.
That's the piece is, you know,not operating in the dark and
then understanding, well, Ithink it's for me, and I think
it aligns with with Alex aswell. It's less about remote

(49:28):
diagnostics, I think again, asnomenclature, and it's more
about, well, which problemsreally warrant a truck to roll?

Yeah, it doesn't, I think we're moving away from, I
think historically, there aretwo types of projects. There are
projects that got a truck rollfor every single fault code that
any inverter through, and thosewere all underwater, and there
were the projects that got no Oand M whatsoever. And you know,
those may have actually beenmore profitable, even though
they were finding out theirissues three months later. So.
Yeah, and the right answer is,like, somewhere in the middle to

(50:03):
just, we're using the wordsfault and alert. When, when
Wattch. We say fault. We mean apiece of hardware, usually an
inverter, but, but also any kindof device, sort of self
reporting an issue, fan, fault,arc, fault, things like that.
For a long time, that was howpeople looked at the health of a
site. Is by tracking downfaults. A lot of tools create an
alert, which is a notificationfor every fault That's crazy. A

(50:26):
lot of these sites generate1000s of faults a month, weaving
of it sort of completelybackwards, which is, let's
actually analyze the performancefrom a first principles
perspective. And this issomething we share with Omni.
And then we can go look at faultcodes to help understand why.
But if you know you have a faultcode that isn't going to you
know, there's also this idea of,like preventative maintenance,
if, if a fault code indicatesthat something is going to die,

(50:49):
that's also important to know.
But we should not be spendinghuman time, human capital, on
every, every single little faultcode. It's it's about creating
the right alerting rule set sothat you only get alerts for
high signal events.

So as asset owners and developers are adding more
storage into their solarprojects, either initially or
retrofitting in storage. Doesthat change the game at all?
From a data collection point ofview and or the the way that

(51:24):
data is collected from thesesites.

Yeah, batteries.
Love batteries. I will, I willthrow one from ninium First,
which is that this, this idea ofavailability versus performance.
It gets much worse withbatteries, because it's, it's a
battery can be available for 95%of the year and only make 5% of
the revenue. Like revenue, likesome of these batteries, the
leverage on that peak demandevent. Or, you know, you know,
if you miss one sort of demandspike for a behind the meter

(51:53):
battery, you can be underwaterfor the whole month of the whole
year. So this, this likeseparation between availability
and actual sort of financialperformance just gets wider. So
you have to, you have to takethe same lessons and then do
even better in the storageworld. Would you agree with
that? Trish, my characterizationthere 1,000%

Yeah, that if that battery isn't performing when
you need it on, on, whetheryou're, you know, following a
curve demand or whatnot, theentire project is is upside
down. Yes, you need it to beavailable performing. You need
to be performing ready to rollwhen those demand spikes go up.

So, yeah, I think storage from a we talk about
solar plus storage there,they're getting married
together. But the when we talkabout the health of a battery.
It's very different than thehealth of a of a solar farm. In
some senses, it's actuallyeasier to measure if you charge
the thing up all the way andempty it and it charges and
discharges at the right rate tothe right total capacity. That's

(52:55):
a pretty good assessment ofhealth. The weather doesn't
really play into it in terms of,sort of the electrical health of
the system. What makes it sodifficult in solar is divorcing
that the weather piece andunderstanding sort of the
weather adjusted. Has it justbeen a cloudy month, or is my do
I have problems? I think toanswer your actual question

(53:15):
about data collection, yes, Ithink right now, most solar
monitoring tools will pull in adata point or two from storage.
They'll give you a meter orthey'll give you a top level
state of charge, but they'rejust not designed for a truly
sort of multi asset to be amulti asset solution. Wattch had
the advantage of being arelatively new business in the

(53:35):
monitoring space. And we, fromday one, said, this is a multi
asset product, and sort ofbatteries have been baked into
our data model from from thevery beginning. And being able
to pull there's also way moredata from an average energy
storage system than there isfrom a solar system. So being
ready to capture all of that andpresent it in a unified and

(53:55):
intelligent way, incrediblyimportant. And it's something
that I think is going to be abig wake up call in the industry
as people start wanting higherstorage attachment rates.
Residential does not have thisproblem, whatever it's worth,
because it's all these likewalled garden ecosystems, but
the larger IPPs won't want to betrapped in any one OEMs sort of

(54:15):
software ecosystem, so they'llhave to be open data standards,
and there are, but getting allthat information into one place
today is a huge gap, and I thinkit's something that we're able
to solve well for our customers,and the demand for that is is
really, really only increasing,and it's one of the things I
like talking to the Omnidianfolks the most about is like,

(54:36):
where do you see this? Becausethink solar is still growing up
compared to wind. I mean,storage, I feel like, is so
nascent in terms of, terms ofthese, these kinds of questions.
So it's exciting to work withother people that are thinking
hard about what the future oneyear, five years, 10 years out
is going to look like.

We've got one more question from the audience that
we should get to, and we're downto the wire. Here, but so thank
you all for being here. Reallyappreciate your participation.
How do asset owners measure thecosts and benefits of third
party O and M? How do they makethat decision whether or not to

(55:18):
hire an Omnidian or to conductall that activity themselves,

to me are, it's, it's, it's not a complex
formula. It really depends wehave, we Alex knows, and I know
we know several IPPs that selfperform. But what, what makes
that business case work is theyhave clusters of DG portfolios
near their utility scaleportfolios. So they've been

(55:47):
really strategic. If they'regoing to self perform, they've
been really strategic about thegeography. We can use our team
that's at this 150 megawatt siteto go also service these other
50 megawatts within 100 mileradius. A lot of it comes down
to, I mean, it's head count. Soone is the technicians,

(56:09):
depending on your portfolio. If,to me, it's pretty simple, it's
going to should be geographydriven, because you, as we
talked about very early on, ifyou have an in house team of,
say, 1020, even 30 technicians.
Are you going to set up regionalhubs? You could, you might,
right. You might have a regionalhub in the, you know, northeast,

(56:30):
a regional hub in the in thesouthwest, to take care of those
assets. What we're seeing withdistributed generation, what we
have seen, what we continue tosee, is scattered portfolios. It
typically doesn't pencil out tohave your your technicians in
house. You may have technicalasset managers in house, and

(56:51):
they are, they are true partnersof ours as well that you you
have folks in in house that areresponsible for the performance
of those assets. But a very Imight be over, overshooting
here, but the only time I haveseen a an asset owner self
perform is when they have had alarge enough portfolio in a very

(57:13):
condensed region to justifyhaving their own technicians.

I mean, this is a little outside of our direct
scope, but I completely agreewith everything Trish said. I
think it's an opportunity costof where you want to develop,
like I think it's, it's, youknow, maybe self performing. You
can make a case that it's lessexpensive, but you are trading
this opportunity cost of sort ofbeing constrained
geographically. And I thinkdifferent business strategies,

(57:44):
and we'll see in the long run, Ithink they'll probably be
successful people doing both.
There are, we are, well,hopefully

we've left, we've left our audience with with some
additional questions, and Iwould like the panelists to
please let our audience know.
How can the audience andlisteners to this recording find
you online or in the real world?
Trish you want to go first

sure Trish Graf at Omnidian, and my email is T
Graf, G, R, A, f@omnidian.comprobably leave it somewhere in
the chat, so I will put that inthere. Just Nussey was quicker
to it.

I just dropped you in there. Did I get that right?
That's my email address. I willmake the same plug I did at the
end of podcast, Tim, which isthat, you know, we say we're
connecting tomorrow's grid. Ialso love connecting with the
people that make tomorrow'sgrids work. We are a very sort
of feedback driven productorganization. If you have ideas,

(58:44):
thoughts, criticism, whatever itis, I would love to hear it and
have a conversation with youabout it. I think everyone
talking about how to build abetter renewables industry is
the way that we move the ballforward. So albeit, SPI, come
find me or sorry. Re plus, andthat's my email. Shoot us. Reach
out. Love to have theconversation and hear what
you're thinking about the space.

Likewise, the entire Omnidian team will be at
re plus, please look out andfind us

great. And you can check out all of our content at
Cleanpowerhour.com Give us arating and a review on Apple or
Spotify, and reach out to me onLinkedIn. I love hearing from my
listeners. Or you can contact mevia the website. I want to thank
Trish Graf with Omnidian andAlex Nussey With Wattch for
joining us for this webinartoday. I'm Tim Montague, let's

(59:33):
grow solar and storage. Takecare, everybody. Thanks, Tim.
Cheers. You.

All Episodes

Data Driven O&M Strategies with Wattch and Omnidian | EP232

Episode Transcript

Popular Podcasts

Stuff You Should Know

24/7 News: The Latest

Crime Junkie

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}Data Driven O&M Strategies with Wattch and Omnidian | EP232