<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Howard Low on Medium]]></title>
        <description><![CDATA[Stories by Howard Low on Medium]]></description>
        <link>https://medium.com/@howardlhw?source=rss-7e557f15bf87------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*RFEgD8L2hamyCRTn7ZLZRA.jpeg</url>
            <title>Stories by Howard Low on Medium</title>
            <link>https://medium.com/@howardlhw?source=rss-7e557f15bf87------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 09:35:36 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@howardlhw/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[AI isn’t making hiring easier, it simply added a layer of unnecessary complication]]></title>
            <link>https://howardlhw.medium.com/ai-isnt-necessarily-hiring-easier-it-simply-added-a-layer-of-unnecessary-complication-746fa73340b8?source=rss-7e557f15bf87------2</link>
            <guid isPermaLink="false">https://medium.com/p/746fa73340b8</guid>
            <category><![CDATA[hiring]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[recruitment]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Howard Low]]></dc:creator>
            <pubDate>Thu, 30 Apr 2026 02:39:38 GMT</pubDate>
            <atom:updated>2026-04-30T02:42:02.085Z</atom:updated>
            <content:encoded><![CDATA[<p>Recently, I have started hiring people to expand the team and scale our products. Peers in the sector would know that most companies are going through tough times due to AI — resulting on massive layoffs across all industries.</p><p>META recently 8000 employees (20 April 2026), cutting global workforce by 10%. This is a global trend across all industries from creative, software, digital marketing, and industries. <a href="https://layoffs.fyi/">https://layoffs.fyi/</a> indicates 18 publicly announced major layoff in April 2026 alone.</p><p><strong>I ran a human experiment, turned out that less than 13% of candidates read through the job description when applying for jobs. Less than 1% of middle management hiring read through the job description.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/687/1*LlJQBH-zaRCePNl2TZH4qQ.png" /><figcaption>Statistics, because in numbers we trust</figcaption></figure><p>The experiment is simple — the instruction for job application is provided at the bottom of the page — which involve answering 5 simple questions by sending an email directly to me. AI Agent can’t do it on their platform and would simply ignore this. If I receive an email from you — you are guaranteed a response from me.</p><p>If you don’t review the job description, it’s only fair that I won’t read your resume.</p><p>AI for recruitment further divides the employee-employer relationship. In 2023/2024, AI was used as response generation tool that provides a one-click solution for applicants to apply to companies. As a result, responses are no longer genuine and we see cookie-cutter response in from applicants like “why do you want to join our organisation”.</p><blockquote>From the hiring standpoint, it is obvious if the candidate uses AI for application. Those usually get filtered out immediately.</blockquote><p>Moving towards 2025, companies started using AI to filter candidates based on job description. The counter-battle from employees is to auto-generate custom-made resume for the specific job to pass through the AI filter. For the first time in the history of man kind, human is no longer involved in the first line of candidate filtering.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/618/1*JtyJlCyw4isx08W-moIqLQ.jpeg" /></figure><p>Lots of factors have contributed to this. In the past, each company has their own Human Resource Management tools that requires candidate to go through lengthy application (1–2 hour per application). Most of the time, these efforts only provide the candidate an auto-reject email 5 minutes after. Then the innovation of “Easy Apply” on Linkedin that makes it easy to apply, but inviting loads of unqualified candidates.</p><p>Job hunting became a number game for candidates. If 100 job application is required to secure 1 interview, you need to get 300 job applications per day to get 3 interviews.</p><blockquote>Job opening on LinkedIn can receive up to 100 candidates in just the first hour of posting, with VP level jobs receiving up to 900 applicants.</blockquote><p>This approach created a big problem for the hiring manager to differentiate qualified candidates. The solution is simple — interview as many candidate as possible — if you have infinite time on hand. The next solution to this is simple — automate the interview process. Dystopian, but <a href="https://news.sky.com/story/british-job-hunters-turning-up-to-find-their-interviewer-is-an-ai-robot-13531971">AI interview </a>is a straight forward solution for employer — automate the interview process to gather employee data through a structured process. Employees counter it with AI interview support tools that makes it impossible even for the hiring manager to notice, even in the even of screen sharing (such as Parakeet AI). This is just the state of recruitment AI tug-of-war between employee and employer in 2026.</p><p>The issues of AI are far more detrimental for fresh graduates. Just as smartphones took away our ability to memorise phone numbers, AI took away their ability to think. I worked with students from universities and I am impressed by the technical aptitude of these students in using AI. During project interview, the students would present beautiful graphics with thorough analysis of the problems given — except that the solution and presentation were AI generated. They present everything word-for-word, but failed terribly at answering some very basic questions that should have been within their curriculum.</p><p>Similar to Tik Tok, AI seems to be driving people for instant gratification of results. It would certainly help deliver results easier but it makes people less patient of waiting for an outcome. After all, why spend 1 week in market research, reading through pages of length reports, when Claude can run a deep research for you in Cowork in a few hours?</p><blockquote>AI tools at the disposal of university students didn’t make them better at performing their future jobs, but quickly transitioned them into middle man project manager — passing information from boss to prompt.</blockquote><p>On the other hand, I see students without strong knowledge of AI falling behind on using AI to maximise the input efforts to achieve given goals.</p><p>The dichotomy of technical literacy in usage of AI will only further divide the workforce in the future. This threatens the availability of technical workforce required to support the economic growth of a country.</p><p>For my company, we always filter candidates manually even though it took us a long time. Shortlisted candidates are invited for a 15 minutes screening interview to know if they are human, followed by multiple rounds of assessment. We highly encourage the applicants to use AI as part of the competency assessment as this is the new reality, but they must be able to justify the why. Most candidates failed at “Why”.</p><p>For the struggling hiring mangers, I highly encourage having a few “standard questions” and “standard responses” that you prompted with AI in advance. If the applicant’s response somewhat resembles the standard response, you know that they are using AI to answer the questions.</p><p>For the struggling job applicants, I highly encourage you to ditch job application via standard job portals — they do not work in your favour. Join networking events, join conferences (plenty of free ones), directly connect with the employees on Linkedin, will dramatically increase your chances of getting hired. It’s not about the quantity, it’s about the quality.</p><p>P/S: For recruiters — We had bad experience with recruiters and will be hiring directly for the time being. Please do not connect with me on LinkedIn</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=746fa73340b8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Danger of Digital Identity Check]]></title>
            <link>https://howardlhw.medium.com/the-danger-of-digital-identity-check-adc407ed5148?source=rss-7e557f15bf87------2</link>
            <guid isPermaLink="false">https://medium.com/p/adc407ed5148</guid>
            <dc:creator><![CDATA[Howard Low]]></dc:creator>
            <pubDate>Wed, 30 Jul 2025 12:08:33 GMT</pubDate>
            <atom:updated>2025-07-30T12:40:24.084Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/665/1*uPfTLK30THzO6xeV_Tr0LQ.png" /></figure><p>Recently, the UK has implemented UK Online Safety Act, which enforces companies to verify their users before granting them access to sensitive contents. This has affected Discord, Reddit, Wikipedia, and among others. While the law has good intentions, certain dangers pertaining to sharing of personally identifiable data is not apparent.</p><p>I will skip through the widely discussed privacy concerns such as data leaks from private companies (Sony is a frequent offender), sharing of personal information, or exposing your identity to potentially malicious actors. I will also not talk about how this can be used politically.</p><p>Let’s think about how the malicious actors can abuse this. Please note that this is not an idea bank for abusing personal information. I’m merely highlighting the dangers it possesses.</p><p><strong>These my opinions. Please reach out to correct me if the facts are incorrect.</strong></p><p>This is LinkedIn’s policy statement when I print screened it on 30 July 2025</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/952/1*wiG3IRNF80V-X_rnZQ4lPQ.png" /><figcaption><a href="https://www.linkedin.com/help/linkedin/answer/a1631613">https://www.linkedin.com/help/linkedin/answer/a1631613</a></figcaption></figure><p>In short, LinkedIn would keep every other data except for those mentioned above (because they are useless to them).</p><p>X (formerly Twitter) is equally vague on their policy as well, as shown below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/722/1*pZ-M1lbgd0sD6FGTUm4L6A.png" /><figcaption><a href="https://help.x.com/en/rules-and-policies/age-assurance">https://help.x.com/en/rules-and-policies/age-assurance</a></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*UCbbReWq7o4B5iWxF5BdRg.png" /><figcaption><a href="https://help.x.com/en/rules-and-policies/age-assurance">https://help.x.com/en/rules-and-policies/age-assurance</a></figcaption></figure><p>In short, if you verify your identity using Persona with X, the information on your ID (which could be passport) will be given to X. Persona will delete the information but not necessarily X. X will learn where you live and share your facial ID with third party.</p><p>This is a very common trick of the social media website to say that they “do not directly retain this data in connection with the User Experience Enhancement”, while keeping silent for data used by other services. This is like saying “we have delete all your photos on our cloud server”, but somehow keep a copy of all your data on a local computer. They aren’t telling a lie, just not the whole truth.</p><p>To have “deleted the information” is insufficient. In images processing, the image itself is useless after we have extract the necessary information from it. Deleting the extracted information is where we can ensure safety.</p><p>In short, once you have shared the ID to a private company, expect the information to be kept perpetually and made available on the black market.</p><p>I have used Persona for user authentication and therefore, fairly familiar with the verification process and their Dashboard. I’m also used their API and this is how the UI looks like from the perspective of X / LinkedIn.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sbgIcbxHO9TzIcQ-4BMjcg.png" /></figure><p>Depending on the type of verification required, you are giving away the following information from the verification process which you don’t provide them at the point of account registration:</p><ul><li>Photo of Photo ID</li><li>Photo taken from phone’s camera (multiples for retries)</li><li>Date of birth and Age</li><li>Home Address</li><li>Country of Birth</li><li>Nationality</li><li>ID Card number</li><li>Signature (from UK Driving License)</li><li>Time it takes to complete the verification process</li></ul><p>The API document of Persona can be found here, which explains what LinkedIn / X can get. — <a href="https://docs.withpersona.com/api-reference/documents/retrieve-a-generic-document">https://docs.withpersona.com/api-reference/documents/retrieve-a-generic-document</a></p><p>It is worth noting that while performing image verification check via Persona (when you take a picture of yourself), that image gets shared with LinkedIn/X. In my experience, most users don’t take picture from a blank background, leading to potential deduction of financial, social, and family background information from the image itself.</p><p>A simple image of a road may be enough for someone to guess the country it is in, in accuracy of several hundred kilometers. Check out the geo-guessers community <a href="https://www.youtube.com/shorts/XGLHLNm_0rw">https://www.youtube.com/shorts/XGLHLNm_0rw</a>.</p><p>Now, think about how much information you can reveal about yourself with a background of your living room.</p><p>Let’s get some examples of the same person that I’ve found on Google Image search. From your selfie verification (and sometimes, re-verification), can you determine if someone is a drug user.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/595/1*t9P4de1S2IacICi0cbEgLQ.png" /><figcaption><a href="https://modernrecoveryservices.com/conditions/addiction/substances/opioids/heroin/before-and-after-addiction/">https://modernrecoveryservices.com/conditions/addiction/substances/opioids/heroin/before-and-after-addiction/</a></figcaption></figure><p>You could determine someone’s wealth from their selfie. Look at the hairline and the amount of work put into personal styling. Skin color reflection can tell you if appropriate skincare has been applied, and eye bags can tell you if the person has adequate sleep (let’s face it, poor people like us don’t sleep well).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/672/1*_z8BwKZusWF5tCwW9cDOTg.png" /><figcaption><a href="https://pagesix.com/2018/07/25/its-highly-likely-elon-musk-spent-over-20k-on-hair-transplant-surgery-doctor-says/">https://pagesix.com/2018/07/25/its-highly-likely-elon-musk-spent-over-20k-on-hair-transplant-surgery-doctor-says/</a></figcaption></figure><p>You can further imply the “before and after” status of the picture by treating the photo on the ID as “before” and the actual selfie as “after”. In the following example, you can tell that this guy has been working out, gotten contact lenses, and is currently working on self image.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/587/1*DLJxFrR-0wPvPnAdhuu-zA.png" /><figcaption><a href="https://www.instagram.com/p/CMnGJPllgNP/">https://www.instagram.com/p/CMnGJPllgNP/</a></figcaption></figure><p>Let’s take a few other pictures from a simple Google search and show you the danger of showing a photo verification. Check if you can identify the following using your own biases.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jGhWp_-doO-QuuzLiwzYLA.png" /></figure><ul><li>From the furniture in the house, can you tell who is wealthy?</li><li>What can you tell about them from the clothing they are wearing</li><li>What can you tell from the house decorations, the type of light used, the natural lighting of the room</li><li>What can you tell about their personality from personal hair/beard styling?</li><li>What can you tell from the color of the wall and wall decorations?</li><li>What can you tell about their social economic status / mindset from the type of phone they used?</li><li>What can you tell from the tilt of their head?</li><li>Threading into dangerous realms, politically, are they leaning, left, right or centre?</li></ul><p>With more selfies from their neighbours, we can further deduce this</p><ul><li>The relative wealth status of the individual</li><li>Culture of society — what the people around there are like</li><li>Political affiliation</li></ul><p>Now the fun part. What can LinkedIn/X do with this data?</p><p><strong>Racial, social and financial background profiling becomes possible with this data</strong>. It is a known fact that a simple postal code can be used to predict the financial wealth status of the individual.</p><p>With sufficient data, it is possible to know the number of members of a household, further indicating financial wealth of the user. If a user lives in a house with 3 other flat mate in outside London versus a user living in a studio in central London.</p><p><strong>Nationality based services / echo-chambers to increase user engagement</strong>. If you’re an immigrant moving to the UK? X and LinkedIn can recommend visa application services to you, specifically. Potentially swinging right? Let’s feed you more with news content that makes you vote right.</p><p>Cross mashing with other information that they already have on you — like your career, interests, interactions with other users, etc — would only give them a better profile of who you really are.</p><p>That is on top of others things you’re familiar with, like targeted ads, targeted engagement, and content suggestions.</p><p>I’m less worried about them selling products but more worried about the profile which they would have created about us from this information that was previously inaccessible to them.</p><blockquote>What if, the data is leaked? The damage caused by this law could be devastating.</blockquote><p><strong>If the exact address as indicated on the Passport and ID has been leaked, this provides perpetrator a home address of their targets without doxxing</strong>. This puts the individual journalist, protestors, politicians, CEOs, and any person of interest in immediate danger. This includes their family members locally and overseas.</p><p><strong>Securities currently in place for user verification will be breached</strong>. With date of birth and passport information, the perpetrator can guess the answer to “where is your hometown” and gain access into any of your non-2FA websites. Or pay a visit to your house at hometown to find out your mother maiden’s name, or primary school, or first car through your family members.</p><p><strong>Financial fraud will be rampant. </strong>The perpetrator can now open a bank account / apply for credit card / apply for loans using this stolen identity.</p><p>Obviously the next argument is that not all websites will be affected by this. When the first hobby site gets fined by the government for non-compliant, every other company would use the verification services to avoid the fines. Let’s get into the details of the Act, as shown below</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/842/1*T3XagDxdsHoCO6fOzj-ErQ.png" /><figcaption><a href="https://www.gov.uk/government/publications/online-safety-act-explainer/online-safety-act-explainer">https://www.gov.uk/government/publications/online-safety-act-explainer/online-safety-act-explainer</a></figcaption></figure><p>I can start a cooking community page online, where user can recommend cooking techniques and tools they used for cooking — including knives. Now I’ll have to verify all my users before they have access to my basic knife skills video that I have posted on the company blog.</p><p>This is where the danger lies, a small startup not likely to understand cybersecurity requirements to protect the user information. A malicious actor just have to browse through the internet and attack the weakest websites with almost zero cybersecurity protection to gain access to personally identifiable information.</p><p>I don’t believe that companies should get more information from you than is needed for them to render the services, especially if we have been paying for their services.</p><p>While all these may look far fetched, remember that Cambridge Analytica changed the outcome of multiple elections using just the data from Facebook posts.</p><p>This is why my LinkedIn account remain unverified.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=adc407ed5148" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Automate Emails Labelling Using AI]]></title>
            <link>https://howardlhw.medium.com/automate-emails-labelling-using-ai-eddc4a880509?source=rss-7e557f15bf87------2</link>
            <guid isPermaLink="false">https://medium.com/p/eddc4a880509</guid>
            <dc:creator><![CDATA[Howard Low]]></dc:creator>
            <pubDate>Wed, 07 May 2025 23:23:48 GMT</pubDate>
            <atom:updated>2025-05-07T23:23:48.017Z</atom:updated>
            <content:encoded><![CDATA[<p>The story on how I started using AI to manage my emails.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*IjctisAww1B-hnXP" /><figcaption>Photo by <a href="https://unsplash.com/@hostreviews?utm_source=medium&amp;utm_medium=referral">Stephen Phillips - Hostreviews.co.uk</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>AI is all the in hype now and it’s essential to separate the hype from the practicality. I tend not to follow the hype and dive more in detail into the practicality aspect of it. Namely, can AI make my life better, if yes, by how much?</p><p>I run an extremely lean startup considering the scope of the work which we delivering. As prioritisation goes, I started spending more time on engineering and less time on email housekeeping. I’ve surveyed my peers across different functional roles only to discover that this is a common problem.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/406/1*ZKSpGn715E7kCvRG1yTRig.png" /><figcaption>The ever increasing unread messages in inbox.</figcaption></figure><p>This is my personal preference, but I do want to keep my emails properly labelled to properly prioritise my attention. Google’s automated filter-based labelling could work, but it’s not foolproof and scalable in labelling the emails.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/409/1*SHeg76ZOX8E7ChoriMKzZQ.png" /><figcaption>The labels that I have created to label the emails.</figcaption></figure><p>The solution is to use n8n with LLM models to automatically label the emails based on its content. To achieve this, I’ve installed a local version of n8n to run a locally hosted docker container. Since n8n is a workflow automation tool, the required computation power is quite minimal. You can build your own AI gents (more on that in the future) to automate tasks like email response automation.</p><p>For those who thinks that it’s easy and it’s just a “drag and drop”. IT IS NOT. The workflow and the API integration means that some form of engineering knowledge is required before starting out. Even designing the workflow is an engineering thinking by itself (consider situations where the workflow will fail, so what would you do). Structured output requires technical knowledge such as understanding of json format, and what is data types (string, boolean, etc). These knowledge aren’t as common as your expected. If you want to know how it looks like, try connecting data input from Typeform, process the data using Zapier, before sending the data to AirTable. This would be a good starting point before diving into n8n.</p><p>I also happen to have a 32GB laptop with a non-defective Intel i7 processor (IYKYK), so this runs perfectly well on the background on my Lenovo Carbon laptop.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HX0DEnWEM75QdwR8Z0V9Bg.png" /><figcaption>Snapshot of the workflow I have created.</figcaption></figure><p>The process of attaching the language model to the LLM chain is easy, you can literally pick any language model you want (OpenAI, DeekSeek, etc).</p><p>The hard part is coding the prompt that generates the desired output. In this case, I have to programme the prompt such that it auto-labels each emails based on the subject, sender and content based on the pre-defined labels that I have created. No AI tool is perfect, and there will always be outliers.</p><p>Integrating with Gmail is a straightforward process. N8n uses OAuth for authentication, which the credentials can be created via GCP.</p><p>The next part is to determine all the possible scenarios, which we have encountered a few odd scenarios and handled them accordingly. This is to make sure that the application would not be trapped in infinite loop which will end up with me paying the LLM providers without achieving anything.</p><p>This is a relatively simple tutorial application (yes, I treated this as a tutorial application because I need it) and the outcome is remarkable astounding.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tiweFI2VzKsuggv7kOBjJQ.gif" /><figcaption>The emails get automatically labelled in the background.</figcaption></figure><p>The question is, is it wort it?</p><p>Absolutely. I spent about 4 hours setting it up from scratch which would otherwise take me much longer just to filter and sort out the emails. With the current workflow in place, I can easily re-configure the labeling for other processes.</p><p>The cost is something I want to talk about. OpenAI’s model costs me 0.2468 cent / API call where as Deekseek costs 0.08368 cent / API call. This is a difference of price by 76%!! Without evaluating the reasoning models of both providers, the real question is whether the model achieve the intended output at the price provided.</p><p>If Deepseek has been misclassifying my emails, then despite the low cost, it would still not be worth the price. For a simple task like this, DeekSeep and OpenAI performed equally well. This makes product differentiation much harder and the stickiness to stay with OpenAI much lower.</p><p>This is a very valuable lessons for startups founders, it doesn’t matter how good is your product, if your customer doesn’t see value in the product that you have provided, your product has no value. If your product, despite great value, has a competitor that allows customer to achieve the same outcome with lower price, product superiority doesn’t matter as much.</p><p>Out of curiosity, I have also build an email agent to automatically generate response based on the content of the email. That’s just an initial workflow I have created to automate my email response — more to come when I have more time to play around in July.</p><p>Obviously the next step is to bring it to production for my company, but that is the story for another day, big plans ahead.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=eddc4a880509" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Demystifying Electricity Tariff in Singapore]]></title>
            <link>https://howardlhw.medium.com/demystifying-electricity-tariff-in-singapore-b81cbeb60f95?source=rss-7e557f15bf87------2</link>
            <guid isPermaLink="false">https://medium.com/p/b81cbeb60f95</guid>
            <category><![CDATA[power-generation]]></category>
            <category><![CDATA[singapore]]></category>
            <category><![CDATA[singapore-power]]></category>
            <category><![CDATA[enverse]]></category>
            <category><![CDATA[electricity-tariff]]></category>
            <dc:creator><![CDATA[Howard Low]]></dc:creator>
            <pubDate>Tue, 27 Oct 2020 16:47:10 GMT</pubDate>
            <atom:updated>2020-10-27T16:47:10.684Z</atom:updated>
            <content:encoded><![CDATA[<p>The upward adjustment of electricity tariff in Singapore has sparked a slight debate within the community in Singapore. During the challenging COVID-19 times, it is understandable price hike is a sensitive topic. Here I try to use my knowledge, to the best of my ability to explain the facts using data.</p><p>We see debates arguing that the price hike is entirely caused by power generators, and that Singapore Power does not profit directly from the price hike. How much of it is true? We’ll look into it using facts and data.</p><p>I do not have all the facts. If you have any questions or want to correct me for any mistakes I made in this article, please feel free to reach out to me at howard@enverselab.com.</p><h3>Components of Electricity Tariffs</h3><p>The first thing to consider is the components that make up of electricity tariff. Singapore Power’s website provided a simple breakdown of the electricity prices.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/490/1*wI71S8v0PbjT_utRj6o8Ug.png" /><figcaption><a href="https://www.spgroup.com.sg/what-we-do/billing">https://www.spgroup.com.sg/what-we-do/billing</a></figcaption></figure><p><strong>Market Admin &amp; PSO Fee</strong> — this is the administrative fee paid to support the services for getting you the best deal (quite literally). They are responsible for purchasing the electricity from the wholesale electricity market at lowest price possible. (Fun fact, if I recall correctly, they have 50,000 equations to determine the lowest cost of energy)</p><p><strong>MSS Fee</strong> — This is the fee paid to Singapore Power Services for meter reading. The meter data will be sent to your electricity retailer for billing purposes.</p><p><strong>Network Costs</strong> — This is the fee paid to Singapore PowerGrid to maintain the electrical network in Singapore.</p><p><strong>Energy Costs</strong> — This is the costs of electricity and is subjected to change due to changing oil prices (to be precise, HSFO180). Why oil but not natural gas, since Singapore is generated 95% of energy from natural gas? Because the natural gas prices has direct correlation with HSFO180.</p><p>Note that SP PowerGrid, SP Services are same but not the same. They are of SP Group but they operating independently.</p><h3>What is the market condition?</h3><p>As of the time of writing this article (25 October 2020), the price of USEP hovers around S$ 60–90 /MWh, or 6–9 cent/kWh. USEP is the price you pay to the power generators, this is the primary source of revenue for the power generators.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/899/1*Wvq9vUe34bl_rBvOBTmpxA.png" /></figure><h3>FACT 1 — Electricity prices correlates with oil prices</h3><p>If you have read my previous article (<a href="https://medium.com/@howardlhw/understanding-singapore-electricity-market-risks-in-dept-analysis-using-data-4e3ab369079b">Evaluating Singapore Electricity Market Risks</a>), you would have noticed the following chart</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/657/1*CsaPbiF3G5ZzjkKQ6C9aGQ.png" /><figcaption>Data analytics by Enverse, green is USEP, blue is HSFO180</figcaption></figure><p>Over a one year period, the price of oil correlates with price of wholesale electricity with a 3 month lag. Note that this is the price of wholesale electricity (USEP), not retail electricity. The correlation between these two are strong.</p><p>How to convert the price from wholesale to retail? Simple, think of it as hedging strategy. The retailers would want to protect itself from sudden fluctuation that kills their business (it had happened before in Singapore), so they add a margin of error. In short, they just add a margin to cover the risk of fluctuation.</p><p>How it works is this, the generators buys in natural gas, and generate electricity. They sell this electricity through Market Administrator (Energy Market Company), to the retailers (which can be internal) or Singapore Power.</p><p>It’s also good to note that Singapore Power nor EMA have almost zero control over this pricing? Why almost zero but not zero? That’s because of vesting contract, but that’s a conversation for another day. EMA sets the rules and framework for the power generators to bid into the system. The price is driven by supply and demand.</p><p>It’s also good to know that the power generators may sign long term natural gas contracts with natural gas suppliers. There is also a take-or-pay contact in place, that essential binds the power generators from under-consuming natural gas. Therefore, the price rightfully should not fluctuate that much. But remember the conversation about <a href="https://www.todayonline.com/singapore/plunge-crude-oil-prices-will-not-necessarily-lead-drop-petrol-prices-mti">why the petrol price went up faster than it drops</a>? Maybe the same mechanism is in place, I don’t have an answer to this.</p><h3>FACT 2 — Electricity prices is not going up all the time</h3><p>This is a human psychological problem. When the price goes down, no one bats an eye. When the price goes up, everyone loses their mind. Here is a summary of historical electricity prices over the past 7 years.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/864/1*Ll7pRrfO0ibZY5O0KE7Xjw.png" /><figcaption>Data from Singapore Power</figcaption></figure><p>Let’s compare it against the oil prices for the past 7 years.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/858/1*-0Yg_BrjwQpBLNgYwHGN0w.png" /><figcaption><a href="https://www.macrotrends.net/2516/wti-crude-oil-prices-10-year-daily-chart">https://www.macrotrends.net/2516/wti-crude-oil-prices-10-year-daily-chart</a></figcaption></figure><p>From reading the charts, it’s clear that the electricity price went down slower than it goes back up, but then that’s business. If you run a power generator business, you would have done the same thing, wouldn’t you?</p><p>Again, this has nothing to do with Singapore Power, because Singapore Power does not really own power generation assets. Why does not really? Because they have small scale solar panels installed (unless my understanding is wrong, feel free to correct me).</p><h3>FACT 3 — Electricity retailer always offer a better deal</h3><p>As of 25 Oct 2020, the price of retail electricity tariff (not regulated electricity tariff) is around high 16c/kWh. Not advertising for these companies though. (We’re open to write articles for a fee, feel free to reach out to me).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/889/1*EqwGDXPFJmcZaZ979P04Dg.png" /><figcaption>Source: Open electricity market on 25 Oct 2020, 3.17pm</figcaption></figure><p>You can say that this is the “true” cost of electricity because these are businesses where they have to remain profitable, non-government related, while remaining competitive. Competitive = cheaper price. Monopoly = premium price.</p><p>The extra cost on the regulated tariff is caused by other factors, like vesting contract. I’m not too sure if vesting contract is the major contributing factor, or sole contributing factor.</p><h3>FACT 4 — Power generators are making more revenue, but so is their long term costs</h3><p>No sane power generator will purchase natural gas when the price of oil is low, just to use it to generate electricity immediately. When the price of oil is low, they purchase more natural gas, but bear in mind, that they might have to consider the possibility of lower future demand. The inventory risk is high on them.</p><p>To put things into perspective of masks. Just because the price of masks are high in the market while you get your masks supply cheap, it doesn’t mean that you should not sell it low? Just because the mask supplier can provide you with cheaper masks, doesn’t mean you have to buy lifetime’s supply of masks right? Same logic applies to power generators. This is what we call exercising market power. EMA created vesting contract scheme to prevent this from happening.</p><p>Remember, it’s just business.</p><h3>FACT 5 — Carbon tax has been implemented</h3><p>It’s interesting how we talk about <a href="https://www.straitstimes.com/singapore/environment/environmental-groups-concerned-about-carbon-emissions-associated-with-flights">carbon emission by SIA</a>, <a href="https://www.channelnewsasia.com/news/singapore/environmental-tax-singapore-airlines-covid-19-travel-aviation-13211568">environment tax on SIA</a>, but no one is talking about carbon tax in power generation industry.</p><p>The carbon tax is capped at S$5/tonne CO2, which translates to 0.20425 cent/kWh (Singapore’s Grid Emission Factor is 0.4085 kgCO2/kWh, it’s really low). This is a tax on generation, so we can expect the cost to be priced into USEP. The plan for Singapore is to double/triple the tax by 2030 <a href="https://www.nea.gov.sg/our-services/climate-change-energy-efficiency/climate-change/carbon-tax">according to NEA</a>.</p><p>As far as I know, there have been many suggestions on how the tax should be used. There isn’t seem to be an organization oversight to manage the carbon tax fund. Rightfully, the tax should be utilized to reduce carbon emission in Singapore. NEA has been doing a great job on that, I’m not too sure if this fund is paid directly to them or to another government organization.</p><p>On 27 October 2020, EMA has announced that <a href="https://www.channelnewsasia.com/news/singapore/ema-grants-energy-efficiency-projects-senoko-tuas-power-ytl-13381500">S$ 23 million in grants has been awarded to three power generation companies for energy efficiency projects</a>. Is this a fair amount? Let’s calculate.</p><p>Carbon Tax started in 2019 at S$5/tonne CO2. Singapore’s power consumption is approximately 51 TWh in 2019, with 95% energy coming from natural gas. Total carbon tax collected is</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/421/1*PPDqDtziAJZAvAqsUpa0KQ.png" /></figure><p>I’m sure government is thinking about reinvesting the tax into renewable startup like my company, right? right????</p><p>Now, shall we get to the juicy part?</p><h3>UNKNOWN 1 — What constitutes the cost of energy?</h3><p>The cost of energy for Q4 2020 is 15.53</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/482/1*wdzkmuNDQKphwH7xjQOBbg.png" /><figcaption>Source: SP Group</figcaption></figure><p>If you compare the last 7 days of wholesale electricity price, which is the price where retailer buys electricity from, the price has been ranging around 6–9c/kWh. Factor in ‘just in case of bad things happen and the price went surging up’ prices, we can assume an internal SP Group of 11c/kWh pricing when priced to consumer.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/905/1*odpgR5tktrT9EBcSLJRSPw.png" /><figcaption>Source: EMC, but this graph is produced by Enerlytics.sg</figcaption></figure><p>15.53c/kWh (Q4 2020 tariff) minus 11c/kWh (safety USEP price) is 4.53c/kWh. We know that some of this profit goes back to the power generators as part of vesting contract scheme, this is a scheme by EMA to curb the market power of power generators, which is the result of low electricity cost today.</p><p>Now the question is, who set the price? Notice that SP used the term “regulated tariff”? The short answer is that it follows a certain calculation according to market condition. This is why the regulated tariff follows closely with market condition determined by global forces.</p><p>I’m pretty sure there is a document on EMA website that describes the formula used to calculate the cost of energy. These are public information. I just don’t have the answer to this question.</p><h3>UNKNOWN 2 — Does SP Benefit from Tariff increase directly?</h3><p>We don’t know. I doubt SP benefits from tariff directly. Tariff is not set by Singapore Power, electricity is never generated from Singapore Power.</p><p>According to Singapore Energy Statics 2020, SP’s market share in retail electricity is 20.4%. While it is good to note that not all are transacting on regulated tariff (some might be transacting on wholesale electricity prices), a good proportion of residents are still purchasing electricity at regulated tariff.</p><p>The calculation for electricity tariff should remain the same. Let’s say part of the 4.53c/kWh is the profit by SP, their profit should remain the same regardless of electricity changes. To illustrate this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*VFlQm27CpnbAiJsbrjIiTw.png" /></figure><p>I believe that SP tariff should work out something like this, as a fixed hedging costs, so basically regardless of energy price fluctuation, they would still be protected. I don’t have knowledge on this, I might be wrong.</p><h3>UNKNOWN 3 — Are the power generators making more money?</h3><p>Hard to say, it really depends on their hedging strategy. The price of electricity is largely driven by oil prices. Oil prices and market demand determine the wholesale electricity prices. On a bad day, you can have this.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/695/1*9W-1LzpEH8SooPjk0fYMXA.png" /></figure><p>You’re seeing it right, zero energy costs. Literally, the power generators are selling electricity for free. this happens on last week of June 2020. I don’t recall anyone talking about it, nor the media reporting about it.</p><p>Imagine this, you are a power generator and you sign on a 6 month natural gas supply since February while the prices of gas was high, and you’re supplying electricity at S$ 0 /MWh, how bad are you bleeding?</p><p>A short Google search will give you enough results to determine whether the power generators are profit making or loss making.</p><h3><strong>UNKNOWN 4 — How strongly is the price of electricity correlating with average USEP price?</strong></h3><p>I have the historical data for USEP in the past 4 years, but its’ quite a hassle for me to pull the data from my server. Drop me a message on LinkedIn or email if you think this is interesting. I’ll probably publish another article if there are enough interests on this.</p><p>In short, the correlation is strong. What I do not know, is the strength of the correlation.</p><h3>How about inflation, does that impact SP?</h3><p>This is where it gets interesting. You must first understand what constitutes “Network Costs”. To understand what is network costs, you will need to first understand the bills paid by commercial customers. Below is a bill simulation tool that shows you what rates will you be paying as high tension customer. The rates should be outdated by now (it was last updated around Q1 2019, I think), but the values shouldn’t deviate too far from this. Transmission charges are SP PowerGrid related charges.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xhMk2AxQ2YabAj6ktTN0Yg.png" /><figcaption>Enerlytics.sg bill simulator</figcaption></figure><p>From the past 7 years historical prices, I have filtered out all the SP PowerGrid related costs (contracted capacity charge, uncontracted capacity charge, reactive power charges, summarize into the follow chart.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EOFvOXaTNxp4aTwFFJse0Q.png" /><figcaption>Source: SP Group</figcaption></figure><p>Putting the price into perspective, you’re talking about inflation of 2–2.5% annually for the network costs. As long as your increment hit that amount, it should be alright. Personally, I think this number is alright, although this is still significantly higher than the Singapore’s national inflation rate.</p><p>To put things into perspective, this amount to approximately S$1.50 monthly addition for a S$200 dollar monthly electricity bill. That’s cheaper than my teh-si-kosong peng. But accounting for inflation over the past 7 years, that’s probably more than a cup of bbt…</p><p>I do not have the historical low voltage transmission charges. Can’t comment on that.</p><p>For all we know, SP Group has a pretty guaranteed YoY 2–2.25 revenue growth. This revenue growth should also tie in with the growth of electricity consumption. From what I know, this is not the their only source of revenue. Take for example, if you want to install a high voltage transmission line to your data center, you will need to apply for high voltage electrical connection. The fee schedule can be found on SP’s website.</p><p>Are you logging in into your trading account yet? If not, I have better news for you.</p><h3>What has been the impacts on SP Group?</h3><p>Let’s put the numbers down in perspectives.</p><p>According to Singapore Energy Statistics 2020, SP Services lost 9.7% of retail electricity market share over 3 years. Singapore’s energy consumption in 2019 is 51.7 TWh.</p><p>To put the finances into perspectives, I have extracted SP Group financial report in the past 3 years.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/663/1*zNo353H_-kEAdw-rtVOLYQ.png" /><figcaption>Source: <a href="https://www.spgroup.com.sg/wcm/connect/spgrp/2517ffe6-f553-4c1c-b65d-f36d67e661ea/SP+Annual+Report+2018+Archive.pdf?MOD=AJPERES&amp;CONVERT_TO=url&amp;CACHEID=ROOTWORKSPACE.Z18_M1IEHBK0MOUJ20ABQK7Q593U32-2517ffe6-f553-4c1c-b65d-f36d67e661ea-neeuc-2">https://www.spgroup.com.sg/wcm/connect/spgrp/2517ffe6-f553-4c1c-b65d-f36d67e661ea/SP+Annual+Report+2018+Archive.pdf?MOD=AJPERES&amp;CONVERT_TO=url&amp;CACHEID=ROOTWORKSPACE.Z18_M1IEHBK0MOUJ20ABQK7Q593U32-2517ffe6-f553-4c1c-b65d-f36d67e661ea-neeuc-2</a></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/743/1*u9kqP8OjqTfadHUsf21Z7w.png" /><figcaption>Source: <a href="https://www.spgroup.com.sg/wcm/connect/spgrp/e72f61b1-de2e-4cc4-a4b5-d8b5f8a3add0/Annual+Report+Year+20182019.pdf?MOD=AJPERES&amp;CONVERT_TO=url&amp;CACHEID=ROOTWORKSPACE.Z18_M1IEHBK0MOUJ20ABQK7Q593U32-e72f61b1-de2e-4cc4-a4b5-d8b5f8a3add0-neeuhS3">https://www.spgroup.com.sg/wcm/connect/spgrp/e72f61b1-de2e-4cc4-a4b5-d8b5f8a3add0/Annual+Report+Year+20182019.pdf?MOD=AJPERES&amp;CONVERT_TO=url&amp;CACHEID=ROOTWORKSPACE.Z18_M1IEHBK0MOUJ20ABQK7Q593U32-e72f61b1-de2e-4cc4-a4b5-d8b5f8a3add0-neeuhS3</a></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ufeRa_C92Bnk9yOVxwkiBw.png" /><figcaption>Source: <a href="https://www.spgroup.com.sg/wcm/connect/spgrp/7b94bd63-2737-48f3-8e7d-248a31ebf772/Singapore+Power+Financial+Statements+201920.pdf?MOD=AJPERES">https://www.spgroup.com.sg/wcm/connect/spgrp/7b94bd63-2737-48f3-8e7d-248a31ebf772/Singapore+Power+Financial+Statements+201920.pdf?MOD=AJPERES</a></figcaption></figure><p>Too confusing? Let me summarize for you. The follow chart puts all the values about “Sale of electricity” into perspective.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/841/1*pWLal4_CsrrKWp5s9QuVbw.png" /></figure><p>It might be worth noting that despite declining market share in SP Services, the revenue for “Sale of electricity” remains high. I have included the average cost of electricity for reference (adjusted according to SP Group’s financial year, average across the 4 quarters). Note that the revenue grew despite of reduction of market share. It seems like the price of regulated electricity helped sustain the revenue growth.</p><p>Meanwhile, the revenue for network costs has grown drastically. Rightfully, the revenue of network costs is priced at cost per kWh. Therefore, the growth rate of network costs should be parallel to growth rate in electricity consumption. From the data, the growth rate of revenue definitely outweigh the growth rate of electricity consumption. SP Group might have other revenue within the network costs segment that drives the growth.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/823/1*50rEt9N1Z3EDDKVU7115wQ.png" /></figure><p>The net profit after tax is mouth-watering.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BzH1WI_14TkpE86x-u-_7w.png" /><figcaption>Net profit after tax, source:<a href="https://www.poweringthenation.sg/chairman-message">https://www.poweringthenation.sg/chairman-message</a></figcaption></figure><h3>Here is my analysis</h3><p>Let’s recap what happened in 2018 to 2020.</p><p>In 2018, SP Services lost 3.1% retail market share, but the electricity tariff went up in 2018. This might have resulted in an increase of YoY revenue by 1.89%.</p><p>In 2019, SP Services lost further 4.2% retail market share, but the electricity tariff went up in 2019. This might have resulted in an increase of YoY revenue by 5.68%.</p><p><strong>Do be mindful that correlation is not causation. The question is, how can SP Group has a YoY revenue growth from “Sale of electricity” with declining market share?</strong></p><p>Meanwhile, the revenue for “Use of System Charges and Transportation of Gas” has increased from S$ 1034 million in 2017 to S$ 1493.5 million in 2020, an increase of 44% by annual revenue in just 4 years. According to the report, revenue from “Use of System Charges” is 1234.1 million while “Transportation of Gas” is 259.4 million in 2020. It can be argued that the increased in revenue is caused by increase in energy demand, but I’m pretty sure the electricity demand of Singapore does not grow 44% in 4 years.</p><p>In 2020, SP Services suffered a YoY revenue reduction of 17.43%. Also it’s interesting to note that the financial year ended on 31 March 2020, the moment before the price of oil prices plunged, and also the year with the largest drop in electricity retail market share. Therefore, we can assume that the reduction in revenue is primarily contributed by the loss of market share (4.2%) with negligible impact from electricity tariff changes. If anyone justifies the reduction of revenue using covid-19 and the plunge of oil prices as the excuses, you know what to tell them.</p><p>Judging from the number, I believe that SP Services might potentially profit from increase in electricity tariff indirectly. Again, these are complicated calculations and I do not have the full facts to everything.</p><h3>In Summary</h3><p>From what I have observed over the past years, the price of electricity very strongly correlates with the change of oil prices. While the impact of oil prices to electricity is strong, it is also worth noting that there are many other factors that impact the price of electricity.</p><p>As for whether SP Group directly benefits from the price of electricity increase, we do not know for sure. The number surely would speak for itself. But I do know for sure that SP Group is not responsible for the change in electricity tariff, and neither is the power generation company. A bulk of the profit goes to the power generation company, but that does not mean that the power generation company is in control of the price. There is some math to explain it.</p><p>It would be interesting to see the impact of USEP on regulated tariff, and see the relationship of the historical revenue by the power generation company, USEP, and regulated tariff. From there we can calculate the correlation strength between USEP and regulated tariff. I’ll probably analyze this data some other time in the future, but I’m a bit too busy at the moment.</p><p>Till next time!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b81cbeb60f95" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Covid Cases in Singapore (up to 21 March)]]></title>
            <link>https://howardlhw.medium.com/covid-cases-in-singapore-up-to-21-march-bde06876a898?source=rss-7e557f15bf87------2</link>
            <guid isPermaLink="false">https://medium.com/p/bde06876a898</guid>
            <category><![CDATA[coronavirus]]></category>
            <category><![CDATA[covid19]]></category>
            <category><![CDATA[wuhan-viru]]></category>
            <dc:creator><![CDATA[Howard Low]]></dc:creator>
            <pubDate>Tue, 31 Mar 2020 19:13:57 GMT</pubDate>
            <atom:updated>2020-03-31T19:16:56.949Z</atom:updated>
            <content:encoded><![CDATA[<p>With the data being public, I’ve decided to dwell into the data and see what we can find out about the spread of Covid-19 in Singapore.</p><p>I scrapped the data from CNA (<a href="https://infographics.channelnewsasia.com/covid-19/coronavirus-singapore-clusters.html?cid=covid19_desktop-banner_19022020_cna">https://infographics.channelnewsasia.com/covid-19/coronavirus-singapore-clusters.html?cid=covid19_desktop-banner_19022020_cna</a>). Unfortunately, the data stopped getting updated after 21 March.</p><p>I’ll shed some light for those outside Singapore. All the cases are being tested indiscriminately, which means that the cases presented here are the population size of the Singapore database. I have limited access to the data, and there isn’t much I can do about it.</p><p><strong>Age Distribution</strong></p><p>It shows that the virus does not discriminate according to age. It infects people as young as 6 months old, and as old as 87 years old.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/921/1*oVLLss9vldLpKpIO5jFO0A.png" /></figure><p>If we use the population statistics (<a href="https://www.statista.com/statistics/624913/singapore-population-by-age-group/">https://www.statista.com/statistics/624913/singapore-population-by-age-group/</a>). Visually, it seems like slightly follow the statistics. Which means that the virus do not discriminate, and can infect (and showing symptoms) a younger patient as much as an elderly patient.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/1*e8pd4neQnWZniLcTUJyKIw.png" /></figure><p>Interestingly, assuming equal distribution, the virus should spread to male and female equally.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/481/1*_R0UktEFZwj3P2szNv1Pfg.png" /></figure><p>This could mean that daily life / activities would have higher impact on the possibility of contracting Covid-19.</p><p>This means that staying home is the best course of action from contracting the virus.</p><p>Fun fact: Despite changing the name from coronavirus to Covid-19, CNA still uses “wuhan” in their html code. Technical debt can be really challenging to eliminate.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/337/1*9Q7GT97BZyIPUl0tm-zMAw.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bde06876a898" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Economics of Tesla Solar Roof in Singapore (Part 3)]]></title>
            <link>https://howardlhw.medium.com/economics-of-tesla-solar-roof-in-singapore-part-3-473724d04822?source=rss-7e557f15bf87------2</link>
            <guid isPermaLink="false">https://medium.com/p/473724d04822</guid>
            <category><![CDATA[singapore]]></category>
            <category><![CDATA[solar-energy]]></category>
            <category><![CDATA[tesla]]></category>
            <category><![CDATA[rooftop-solar]]></category>
            <category><![CDATA[solar-roof]]></category>
            <dc:creator><![CDATA[Howard Low]]></dc:creator>
            <pubDate>Thu, 18 Jul 2019 14:31:42 GMT</pubDate>
            <atom:updated>2019-07-18T14:31:42.204Z</atom:updated>
            <content:encoded><![CDATA[<p>This is a continuation of the article from part 1 (<a href="https://medium.com/@howardlhw/economics-of-tesla-solar-roof-in-singapore-part-1-485710f3f62?source=post_page---------------------------">https://medium.com/@howardlhw/economics-of-tesla-solar-roof-in-singapore-part-1-485710f3f62</a>) and part 2 (<a href="https://medium.com/@howardlhw/economics-of-tesla-solar-roof-in-singapore-part-2-863007ab3dca">https://medium.com/@howardlhw/economics-of-tesla-solar-roof-in-singapore-part-2-863007ab3dca</a>). Please read at least part 1 before continue reading.</p><p>The main motivations for residential owners to use renewables are financial motivations or environment motivations. Often, the financial motivations are the key factors driving purchasing decisions. As it was previously discussed, the economics of Tesla solar roof is not feasible. The question is — what can be done to make it more feasible?</p><p><strong>Targeted price</strong>: Considering the hefty upfront investment required, let’s assume that the end-user expects a 2% annual return from the upfront investment, this put a targeted savings of SGD 89,847 over 30 years, or monthly savings according to the following chart (with 2.5% annual inflation)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*B7mEAA-NfXfMq0oKBxuyWg.png" /></figure><p>It’s simple factory economics, to make the solar roof “profitable”, you either increase the revenue (energy generation) or reduce the cost. Note that leveraging finance with the bank isn’t an option since the impending interest rate would be higher.</p><p>Increasing the energy generation isn’t quite possible since it’s limited by technology and weather conditions — where the end-user has no control.</p><p>The only parameter the end-user can adjust is the cost reduction. As a huge bulk of the cost comes from Tesla, a reverse calculation from the targeted saving would indicate the “targeted” selling cost. The breakeven point for Tesla Solar Roof system is SGD 81,009, which is approximately 30% reduction is the quoted price of SGD 115,805.</p><blockquote>Tesla would need to reduce the price for solar roof <strong>system </strong>by at least 30% to make it attractive for the Singapore market.</blockquote><p>As the current consideration is a combined system of solar roof + power wall + roof repairs, the alternative consideration is the usefulness of the power wall and whether it helps in the economics. Let’s start by removing the power wall as part of system cost. That means the targeted solar roof + roof repairs price is SGD 67,367, putting the solar roof price at SGD 52,938. Note that from the quotation, the cost of solar roof is SGD 87,734.</p><blockquote>Tesla would need to reduce the price of solar roof by 40% to be attractive to Singapore market, assuming all energy generated is locally consumed.</blockquote><p>Tesla solar roof is a novel idea in Building Integrated Photovoltaic (BIPV). As an engineer, I totally love the idea behind this. Without getting the economics right, it would be extremely difficult for Tesla to penetrate the solar roof into Singapore market. Granted, Singapore would not be their target market in the first place.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=473724d04822" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Economics of Tesla Solar Roof in Singapore (Part 2)]]></title>
            <link>https://howardlhw.medium.com/economics-of-tesla-solar-roof-in-singapore-part-2-863007ab3dca?source=rss-7e557f15bf87------2</link>
            <guid isPermaLink="false">https://medium.com/p/863007ab3dca</guid>
            <category><![CDATA[tesla]]></category>
            <category><![CDATA[solar-energy]]></category>
            <category><![CDATA[singapore]]></category>
            <category><![CDATA[solar-roof]]></category>
            <dc:creator><![CDATA[Howard Low]]></dc:creator>
            <pubDate>Wed, 17 Jul 2019 15:41:39 GMT</pubDate>
            <atom:updated>2019-07-18T14:32:17.418Z</atom:updated>
            <content:encoded><![CDATA[<p>This is a continuation of the article from part 1 (<a href="https://medium.com/@howardlhw/economics-of-tesla-solar-roof-in-singapore-part-1-485710f3f62">https://medium.com/@howardlhw/economics-of-tesla-solar-roof-in-singapore-part-1-485710f3f62</a>). Please read part 1 before continue reading.</p><p>To make the model more realistic, the following considerations should be made:</p><p><strong>Losses</strong>: tons of them! Every point introduces additional points of losses. Without the exact system in place, it is pretty hard to gauge the losses from the practical standpoint at this stage. When you have thousands of tiles stacked on top of one another. Contact losses is just the beginning of the system performance equations.</p><p><strong>Position of roof top</strong>: not all houses are built facing the sun. If a house is south facing, the question to ask is whether the solar roof should be installed on the north facing roof top. The choice for the buyer would be whether to cover the entire roof top with solar tiles knowing the north facing roof tops would get lower ROI, or to have normal tiles on north facing roof top and solar tiles on south facing roof tops. It becomes a choice between economic and aesthetics. (If the house is built in the southern hemisphere, then it’s different direction).</p><p><strong>Degradation</strong>: The entire calculations did not factor in degradation. Needless to say, it will only make the economics less impressive. Let’s keep it that way.</p><p><strong>Utility of the battery</strong>: The utility of the battery is of questionable value. A landed property in Singapore could have an estimated 7 kW system installed and the actual real generation could be 5.6kW, if you’re lucky, that’s like running 3–4 A/C inverter units at home. Even after deducting the rest load of a residential unit, there will still be exported energy to the grid, and energy export to the grid is bad economics.</p><p><strong>Financing</strong>: The current Fixed Deposit rate in Singapore is approximately 2%. That would transform the initial CAPEX from SGD 115,805 to SGD 205,652 in 30 years. Would the end-user save SGD 89,847 from the solar power system over the coast of 30 years? Or SGD 250 per month average? This is the break-even point. (Note that the average only demonstrate the simple average, it does not account for inflation, etc).</p><p><strong>Maintenance</strong>: The quotation shows the cost of CAPEX, the maintenance cost would likely to occur throughout the lifespan of the product. For starters, inverters only have a 5-year warranty (which goes to tell you the lifespan). If the solar roof is connected in series, a break in one tile would likely cause a big problem in the entire roof — think about the time it takes to “troubleshoot” the roof and the cost involved.</p><p>I really hope that Tesla would iron out all these potential challenges that I could see upfront without even going into the practicality of it. I believe a huge market that they should tap into is the existing rooftop market. It would be really nice if the cost of tiles could be cheaper to make the economics work better, and thus incentivize the adoption of solar energy at the residential level.</p><p>The next part (<a href="https://medium.com/@howardlhw/economics-of-tesla-solar-roof-in-singapore-part-3-473724d04822">https://medium.com/@howardlhw/economics-of-tesla-solar-roof-in-singapore-part-3-473724d04822</a>) will focus purely on the financials and addressing what needs to be done to meet grid parity. Note that it will be number-driven.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=863007ab3dca" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Economics of Tesla Solar Roof in Singapore (Part 1)]]></title>
            <link>https://howardlhw.medium.com/economics-of-tesla-solar-roof-in-singapore-part-1-485710f3f62?source=rss-7e557f15bf87------2</link>
            <guid isPermaLink="false">https://medium.com/p/485710f3f62</guid>
            <category><![CDATA[singapore]]></category>
            <category><![CDATA[solar-roof]]></category>
            <category><![CDATA[rooftop-solar]]></category>
            <category><![CDATA[solar-energy]]></category>
            <category><![CDATA[tesla]]></category>
            <dc:creator><![CDATA[Howard Low]]></dc:creator>
            <pubDate>Sat, 13 Jul 2019 17:01:48 GMT</pubDate>
            <atom:updated>2019-07-18T05:56:25.549Z</atom:updated>
            <content:encoded><![CDATA[<p><em>It’s been a while since I last wrote an article. I have traveled to 4 continents, 8 countries, and 15 cities. That’s almost like one season of Amazing Race!</em></p><p>A report from Electrek (<a href="https://electrek.co/2019/06/14/tesla-solar-roof-quote-price/">https://electrek.co/2019/06/14/tesla-solar-roof-quote-price/</a>) states that Tesla is current completing the third version of its Solar Roof. The real question is — does solar roof makes economic sense in Singapore?</p><p>The total system price is USD 85,314 = SGD 115,805 (on 14 July). The cost is to be recovered by the energy generated from the 9.45kW solar roof top. That accounts to approximately 17 MWh of energy generate per year, generously speaking. Forecasting energy prices 30 years in the future is impossible, no matter how good is the model and how smart is the quant. If there is any indication, the average quarter Singapore Power Tariff is 22.17 cent/kWh since 2014. That is to say, the growth in electricity tariff isn’t really significant, I would generously give a 2.5% annual growth rate in electricity price.</p><p>Another thing that happened to Singapore is the open electricity market (OEM), which gives people an even more affordable electricity prices! Which means that we can assume the price to go from 19 cent / kWh with 2.5% CAGR. Remember, I’m generous in my calculations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*jW9hrxQKfH56_PgQlfZ74Q.png" /><figcaption>2016 to 2019 electricity prices</figcaption></figure><p>Under a very optimistic scenario, if the energy tariff could go up to 38 cent / kWh 30 years later, the solar roof generate an energy value of SGD 143,860. That’s a net profit of SGD 28,055.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/481/1*oxffBrgwOgd-LAKtdiawMQ.png" /></figure><p>Putting things into perspective of financing, instead of investing in a solar roof, the CAPEX is deposited into the bank with 1% annual interest. You would get SGD 154,542 eventually. That’s a net profit of SGD 38,737.</p><p>Obviously putting the money in the bank with 1% Fixed Deposit would give you a better return than investing in solar roof.</p><p>The verdict — installing Tesla Solar Roof is not economically viable. Still, I highly recommend transition into renewable energy because climate change is very real, and we are probably the last generation to take real actions starting from lifestyle choices!</p><p>Part 2 of the article (<a href="https://medium.com/@howardlhw/economics-of-tesla-solar-roof-in-singapore-part-2-863007ab3dca?sk=493a47b0ddfa121b93090dcc023a0038">https://medium.com/@howardlhw/economics-of-tesla-solar-roof-in-singapore-part-2-863007ab3dca?sk=493a47b0ddfa121b93090dcc023a0038</a>) will focus more on getting the estimation more realistic and some adjustments to the model.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=485710f3f62" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Would Grid Battery storage works, financially, in Singapore?]]></title>
            <link>https://howardlhw.medium.com/would-grid-battery-storage-works-financially-in-singapore-771587a10844?source=rss-7e557f15bf87------2</link>
            <guid isPermaLink="false">https://medium.com/p/771587a10844</guid>
            <category><![CDATA[energy]]></category>
            <category><![CDATA[electricity]]></category>
            <category><![CDATA[tesla]]></category>
            <category><![CDATA[power-grid]]></category>
            <category><![CDATA[singapore]]></category>
            <dc:creator><![CDATA[Howard Low]]></dc:creator>
            <pubDate>Sun, 21 Oct 2018 17:34:57 GMT</pubDate>
            <atom:updated>2018-11-02T02:54:35.142Z</atom:updated>
            <content:encoded><![CDATA[<p>Disclaimer: All information / data presented here are publicity available data and of my opinion to the best of my understanding. This does not represent the comment of any companies I have worked for, or with.</p><p>Grid scale battery operates with different grid economics. To understand the economics of grid scale battery, we first must understand the general purposes of grid-scale battery. In general, the two main propositions are to support the grid with a better power quality, or to harness the wholesale electricity price arbitrage.</p><p>In the context of Singapore, the issue of grid power quality is well maintained by Singapore Power. Here, I will focus on the arbitrage of wholesale electricity market. To be more specific, I will try to answer these two questions: (a) does it make commercial sense to install grid-battery storage, and (b) does it make investment sense for consumer as on-site off-peak energy storage.</p><h3><strong>The Battery</strong></h3><p>Money comes first. The following table summarizes the properties of Tesla’s Powerwall 2.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/609/1*7hbLBGyqGzSUBYk8RzftlQ.png" /><figcaption>Tesla Powerpack 2 Technical Specification</figcaption></figure><p>Recently, <a href="https://www.greentechmedia.com/articles/read/tesla-hikes-powerwall-prices?utm_medium=email&amp;utm_source=GridEdge&amp;utm_campaign=GTMGridEdge#gs.Y8===BE">Tesla has adjusted the price</a> of PowerWall to better reflect market values. From the article, it is evident that the cost / kWh used here reflects the extreme lower end of the spectrum.</p><h3><strong>The Container</strong></h3><p>To translate grid-battery into physical space, we will use a standard shipping container dimension. Let’s assume that the grid-scale battery will be shipped in a standard-issued shipping container.</p><p>Assuming that we can pack the batteries up to 80% of the size of the container. This means that a standard container would ship with 13 Tesla Powerpack 2 batteries with the following specification.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/607/1*F3-CDgDoJHfO5MQ1mju5ig.png" /><figcaption>Standard ISO Container Technical Specification</figcaption></figure><p>For the ease of calculations, we presume that any financial projections of the batteries can be met technically. This essentially disregards the technical limitation of the batteries and at its connection point. For example, the extreme case of financial calculation may require a 1000 kWh battery to be discharged in 1 minute, which pushes the export power to 1000 x 60 = 60 MW. This is not possible due to physical engineering limitations. For simplicity, we will presume that such energy discharge is possible.</p><p>We also assume that the battery is 100% efficient. This means that the battery does not suffer any loss in energy during the charging and discharging process.</p><p>Update 2 Nov 2018 — The grid battery used in Singapore’s microgrid system is 200 kWh, at half the volume of the standard container size. So you know I actually oversize the container in an attempt to make it feasible.</p><h3><strong>The Inverter</strong></h3><p>Batteries runs on direct current (DC), while electricity grid runs on alternating current (AC). An inverter is needed to convert DC to AC. Let’s assume that Huawei Inverter is used.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/606/1*ENbSrg-hX37WwhnACSsXUA.png" /><figcaption>Inverter Technical Specification</figcaption></figure><p>We will need approximately 62 inverters to meet the battery output of 2600 kW. This translates to S$ 434,000 in CAPEX. Note that I simplified the engineering assumptions drastically. Life is complicated enough, let’s not make it worse. I got this price off Google but I’ve lost the source. It was from a European B2B e-commerce website.</p><p>This means that it would cost S$ 2 million, minimum, to build a container sized portable battery storage system. Pretty pricey.</p><h3><strong>The Australia Tesla System</strong></h3><p>To validate the costs, we use the <a href="http://mobile.abc.net.au/news/2018-09-27/tesla-battery-cost-revealed-two-years-after-blackout/10310680?pfmredir=sm">Australia grid connected battery</a> as benchmark for battery system costs. Tesla has commissioned a 100 MW grid-scale battery system in Australia. Using the public source of data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/607/1*yoAVdptJx-7ffwNCvIlLEg.png" /></figure><p>The equivalent cost for a 2600 kWh of battery system would be S$ 2,496,000. The cost is approximately half a million higher than our estimates as the we did not include engineering costs, consultancy costs, administrative costs, etc. Not to mention that the notoriously high labour cost in Australia. We now know that our estimate is decently close enough.</p><p>As the container-sized Tesla battery is a 2,600 kWh battery, the linear scale of annual recurring cost would be S$ 110,760–141,085 per year.</p><p>Again, just estimations.</p><h3><strong>The Fundamental of Trading</strong></h3><p>The fundamental of trading is a simple economics concept — buy low sell high. The following chart shows the electricity price over the course of one week. While there are fluctuations from time to time due to various reasons, in general it remains relatively stable, with occasional spikes that doesn’t last long.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/578/1*zS1Ny3Rn1FSta5y_iv6aeA.png" /><figcaption>System Demand (MW) and USEP (S$) from 1st August 2018 to 7 August 2018</figcaption></figure><p>Recently, the USEP has been more erratic.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/578/1*GOQe3wQYILsKFFo7yzWwzA.png" /><figcaption>System Demand (MW) and USEP (S$) from 14 October 2018 to 20 October 2018</figcaption></figure><p><a href="https://medium.com/@howardlhw/the-one-day-of-blackout-in-singapore-gives-solar-system-owners-4x-the-usual-revenue-and-cost-af756853be09">The blackout occurred on 18 September 2018 in Singapore sends the USEP price to all time high of S$ 1354.60</a>. USEP remains relatively high for several weeks after the blackout. Although this is an undesirable condition, it will boost the economics of grid-scale battery.</p><p>In addition, electricity consumers would still be required to pay for transmission and distribution charges at the point of purchase. For illustration purposes, you can refer to this <a href="https://www.serayaenergy.com.sg/contact-us/bill-explanation/">sample bill</a> to understand how the charges are incurred to consumer.</p><p>The <a href="https://www.spgroup.com.sg/wcm/connect/spgrp/dd1fd96d-0c21-4e88-8ded-210ceb61ca3a/%5BInfo%5D+Transmission+Service+Rate+Schedule.pdf?MOD=AJPERES">3rd party charges varies at different voltage level</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/464/1*KSu_Y9r_s8AWQUi-uM8lOA.png" /></figure><p>Grid scale battery is not likely to be connected at 66kV. Therefore, we will use the case of high tension — large transmission charges and low tension charges for simulation.</p><h3><strong>The Price Arbitrage</strong></h3><p>The price arbitrage represents the profit for the purchase and sale of electricity. We will start with a simulation that would be unrealistic — purchase wholesale electricity instantaneously at the lowest price and selling it at the highest price over the 24 hour cycle.</p><p>For the purpose of confidentiality, let’s just imagine a black box that purchases electricity automatically, and it guarantees that the best electricity deal will be made. Obviously, the algorithm cheated the system. If cheating doesn’t help making the business case, we can throw the business case out of the windows.</p><p>Using the actual USEP price from 20 July 2018 to 20 October 2018, the following table shows the transactions that would have taken place. Each row represents a buy-and-sale transaction. Due to the limitation by medium, I’m cropping it piece-wise.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/574/1*P4vxW3BSPLZpcqkrE_VUJw.png" /><figcaption>Transactions in the first month</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/574/1*KrY9LcClbIF9OCFQT3c93Q.png" /><figcaption>Transactions in the second month</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/574/1*rSVcOmotSr0MgDDG2Px-Kg.png" /><figcaption>Transactions in the third month</figcaption></figure><p>I digress: Despite rising USEP, it is still possible to make daily profit if you are doing it right.</p><p>Simplifying the table, the follow chart shows the price arbitrage for every transactions for the past 3 months.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/928/1*Kp-lN8Bx59ngkiGNrO08wQ.png" /><figcaption>October has been a good month.</figcaption></figure><p>Generally, the energy procurement point occurs during off peak and selling point during peak hours. Some exceptions do happen. The following table shows the monthly gross profits. We can see that the first month is the stable month, which is the expected revenue under normal operating circumstances. The second month is the disaster month (due to the profits generated from blackout). The third mouth is post recovery natural disaster month.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/608/1*c1pV7UklDh_49LXbNpYfkw.png" /></figure><p>The sum of transaction demonstrates the arbitrage profits in Singapore dollar per MWh. For a 2,600 kWh, or 2.6 MWh system, the profit is a straight forward calculation. The system would have made a profit of S$ 9,408 / MWh x 2.6 MWh = S$ 24,461 over 3 months. The simple forecasted annual profit would be S$ 97,843.</p><p>This is just the gross profits off the USEP arbitrage, to fully understand the business model, the transaction costs must be accounted for.</p><h3><strong>The Economics</strong></h3><p>There are two scenarios for consideration — grid scale battery arbitrage and on-site energy storage facility. The simulation only considers the trading activity, but neglected the associated transaction costs. The applicable transaction costs include Use of System (UoS) charges and contracted capacity charges. These two scenarios will treat the transaction costs differently and provide different financial outcomes.</p><blockquote>Fun fact: if every month and every year is Spooky October 2018, the annual revenue would be S$192,551. This makes the simple payback of 10 years.</blockquote><p><strong><em>(A)</em></strong><em> </em><strong><em>Grid Scale Battery</em></strong></p><p>Supposed a company is set up with the purpose of making money by buying energy from the grid when the price is low, and selling it back to the grid when the price is high. That means the company needs to ensure that the price arbitrage is higher than transaction costs, and that it would be sufficient to justify its investment.</p><p>The 2600 kWh system would have generated an annual revenue of S$ 97,843. The estimated initial investment cost is S$ 2 million. Hence, the estimated simple payback time is 20 years. This excludes financing cost, annual maintenance costs, operation costs, and many other applicable costs. Without putting in the transaction costs, it is clear that there is simply no business case for any company to operate a grid arbitrage company in Singapore. At least I wouldn’t put my money there.</p><p>With this level of annual revenue, it couldn’t even afford Tesla’s annual maintenance cost. Without the need to factor in the transmission and distribution charges, it is a simple conclusion that a grid arbitrage battery company would file for chapter 11 eventually.</p><p>Fun fact: if every month and every year is Spooky October 2018, the annual revenue would be S$192,551. This makes the simple payback of 10 years.</p><p><strong><em>(B)</em></strong><em> </em><strong><em>Grid Battery as On-Site Energy Source</em></strong></p><p>In this consideration, the grid scale battery is used to purchase energy at lowest price of the day, and consume the energy during peak prices to reduce energy purchase from wholesale electricity. The battery can be either owned by the consumer themselves or third party vendors. The major difference is that the consumer would be paying for lower Use Of System (UoS) charges when purchasing electricity at night and paying lower unit energy costs.</p><p>Here we will consider the case of High Tension Large customer, and a 400V low tension customer. We will start by calculating the savings from UoS. The following table summarizes the differences of two connection types:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/607/1*eZxrAX7pBE5AhJ5WD2Chew.png" /></figure><p>For the high-tension customer, the on-site battery supply can reduce peak demand. The immediate financial benefit would be the reduction of contracted capacity, which is a monthly payable fee to Power System Operator (PSO). The battery system can supply a continuous power of 60kW, which we can safely assume that a 60kW reduction in contracted capacity.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/909/1*9WlIjzcz-dHmDbmvYyvFsQ.png" /></figure><p>Putting everything together:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/608/1*BekOHb4P8xScSKJsEEdguQ.png" /></figure><p>Frankly speaking, I wasn’t expecting the annual savings for HT and LT to be that similar. Nevertheless, we aren’t seeing much difference in terms of savings due to the highly efficient and affordable grid charges.</p><h3><strong>Conclusion</strong></h3><p>Based on the current market arbitrage and the prices of batteries, the simple simulation demonstrate the potential returns from the installation of container-sized grid battery. There are two scenarios for consideration — the first involves the sale of electricity back to the grid, and the second scenario for self consumption. The economics and considerations for both scenarios are clearly defined.</p><blockquote>From the simulation, it is clear that operating a grid battery arbitrage storage company would not make good investment</blockquote><p>From the simulation, it is clear that operating a grid battery arbitrage storage company would not make good investment. Even considering the short term severe wholesale electricity price fluctuation and forecast it to long term, the revenue generated would not be sufficient to pay the installment for the installation of grid battery.</p><p>The business case for individual businesses to purchase electricity at night for peak consumption is feeble at best. Even with the benefit of use of system, the current price arbitrage at wholesale electricity price does not make grid battery system financeable. The returns of the grid-battery storage are highly dependent on the nature of business and the engineering consideration. While the initial numbers looks discouraging, I strongly suggest further investigating this possibility before giving it a complete dismissal.</p><p>One possible investigation is to understand the level of price arbitrage required to make the business case. This requires additional input, such as the O&amp;M costs, IRR, depreciation rate, etc. The estimation would result in a go-no-go grid arbitrage price, that would determine the economics and commercial viability of the grid project.</p><p>Simply put, there are ways to make the business case possible, you just gotta know how.</p><p>I’m certain this article would save some companies valuable time in evaluating the business cases in Singapore. You can thank me by buying me a beer =).</p><p>If you have any comment / questions, you can reach out to me at <a href="/service/mailto:hanwen.low@gmail.com">hanwen.low@gmail.com</a>. If you want to be notified of the next update, you may drop me an email as well.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=771587a10844" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The one day of Blackout in Singapore gives solar system owners 4x the usual revenue, and cost…]]></title>
            <link>https://howardlhw.medium.com/the-one-day-of-blackout-in-singapore-gives-solar-system-owners-4x-the-usual-revenue-and-cost-af756853be09?source=rss-7e557f15bf87------2</link>
            <guid isPermaLink="false">https://medium.com/p/af756853be09</guid>
            <category><![CDATA[energy]]></category>
            <dc:creator><![CDATA[Howard Low]]></dc:creator>
            <pubDate>Sat, 22 Sep 2018 07:40:47 GMT</pubDate>
            <atom:updated>2018-09-22T07:40:47.803Z</atom:updated>
            <content:encoded><![CDATA[<h3>The one day of Blackout in Singapore gives solar system owners 4x the usual revenue, and cost consumers 3x the usual cost</h3><p>Disclaimer: All information / data presented here are publicity available data and of my opinion to the best of my understanding. This does not represent the comment of any companies I have worked for, or with.</p><p>On 18 September 2018, at 1.18am (for the auspicious Chinese, it is 18, 18, 18), Singapore suffered a <a href="https://www.straitstimes.com/singapore/power-trips-in-sembcorp-senoko-units-led-to-outage">massive blackout that leaves 146,797 residents without electricity supply</a>. Electricity supply was restored in just 38 minutes (another auspicious number!). This is by all means, a very impressive feat to achieve.</p><p>I will focus on the impacts of the blackout on <a href="https://goo.gl/x8ajW3">Unified Singapore Electricity Price</a>. The first thing we will look at is the historical market data. The follow charts show the energy market data of the week before the blackout and the week of the blackout.</p><p>The columns of the graphs are separated by days, starting from Friday. The half-hourly demands are plotted on secondary axis. In parallel, you can see the USEP price, primary energy, regulation and contingency price on the left axis.</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*5N6wp7wHTBlrXWFwILHARQ.png" /><figcaption>Market Energy Data on the week before blackout. The ‘baseline’ of the energy data.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*dy9pVIrfk5aOdvuZbv_9Yw.png" /><figcaption>Market Energy Data on the week after blackout.</figcaption></figure><p>There are several observations you can draw from the market data.</p><p>(1) There is no statistical significant change in the energy consumption</p><p>(2) There are several spikes on 18 September 2018.</p><p>(3) The price of primary energy surges on 18 September 2018.</p><p>The small spikes on 15 September and 19 September are of not significant interest at the moment. It occurs from time to time due to various reasons. The pool volatility was <a href="https://medium.com/@howardlhw/turbulent-times-in-electricity-industry-what-you-should-know-as-consumer-b37243c3e84f">high in the past</a> but has remain stable recently.</p><p>We will focus our analysis on the 18 September. The graphs below shows the energy data over the 48 periods.</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*z5Kmi44Fb4TwrkpyvT0PVg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*eyK4kvNaSUN0cfSynAF3RQ.png" /></figure><blockquote>The blackout occurred at 1.18am, which effectively occurs during period 3 (1.00am to 1.30am) , and lasted until period 4 (1.30am-2.00am). Interestingly, USEP remains stable in these two periods. The impact of blackout onto USEP seems to occur at period 5 (2.30am-3.00am), where the price of primary energy surges well above S$ 2000 / MWh, and sent the price of USEP to S$ 924.33 / MWh.</blockquote><p>The price of primary energy seems too high. I’ll be cross validating this value.</p><p>Note that the USEP price for period 1–4 is S$ 172.36 / MWh, S$ 151.48 / MWh, S$ 137 / MWh, S$ 152.90 / MWh respectively. It was already higher than the average USEP price of S$ 110 / MWh. Could this be foreshadowing upcoming incident?</p><p>Over the next few periods, the price of USEP gradually dropped from S$ 924.33 / MWh to S$ 129.90 / MWh in period 12 (05.30am-06.00am). It is an expected behavior as the electricity supply would have been stabilized in over 4 hours.</p><p>The weirdest thing that happened in the next few periods is the surge of electricity price from period 12 onward, up to S$ 1354.60 / MWh in period 22 (10.30am-11.00am). This is the highest USEP recorded in the year of 2018.</p><p>The cause of this is unknown, but this is not the subject of interest here.</p><p>This represent tremendous revenue for solar owners who sells electricity back to the grid! For every kWh of electricity injected to the grid, solar system owners are paid the USEP price (minus other charges, but let’s keep things simple). Let’s assume that the solar system generate electricity with the following generation curve. For convenience, the USEP for 17 Sep and 18 Sep is plotted on the same axis. Note that USEP was stable at approximately S$110/MWh on 17 Sep.</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*zgBQ3kM2iIBxyPaWoz2SMw.png" /></figure><p>The calculation is a simple sum-product of 48 periods. For a 1 MWp solar power system, assuming performance ratio of 1. The system is expected to generate S$ 1180 daily (based on 17 Sept 2018 USEP).</p><p>On 18 September, the solar power system would have generated S$ 4706! That is 300% more than what the system makes normally. However, we should also take into account that the weather is rather cloudy and rainy this week. The actual revenue is expected to be lower.</p><p>On the flip side, it is also worth mentioning the amount that the consumers would have ended up paying. By calculating system demand x 0.5 hour x USEP, Singapore consumers in totality pays S$ 16.87 million dollars for wholesale electricity (excluding transmission charges, etc) on 17 Sep. On 18 Sep, the amount rose to S$ 50.65 million dollars. Refer to table below for periodical comparison.</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*fuNKHV8uDbdKR1NpwSlG5g.png" /></figure><p>The real question is — who would be paying for the extra S$ 33.78 million dollars? I guess we all know the answer.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=af756853be09" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>