A thread about a machine that ran warm from day one, and how the reservoir drawing and the cooler budget disagreed for three weeks before the numbers settled it. Names trimmed.
From: D. Okafor, Customer Commissioning — To: R. Almeida, Design; cc project manager
Subject: New unit HPU-07 oil temperature 62 C on a 22 C shop floor
Ricardo, the press packaged with the 220-litre hydraulic reservoir reached 62 C at the tank surface probe after a full four-hour production run, and it stays there, it does not climb further and it does not come back down. My commissioning acceptance calls for 55 C maximum at 40 C ambient, and we have 22 C ambient and still a 7 degree miss, so I am formally flagging it. The customer runs the machine two shifts, the flow is 60 litres per minute with 4 cylinders on the press cycle, and their maintenance man tells me the old machine from the previous supplier used a much larger tank and never needed a cooler. I have three options to put to you: enlarge the reservoir, add an oil cooler, or reduce the relief setting until the oil stops smoking, and I would rather not do the third.
From: R. Almeida, Design — To: D. Okafor, P. Lindqvist, Tank & Cooling Supplier; cc project manager
Subject: RE: HPU-07 — calculate before we weld
Dami, before anyone welds a taller tank or slips a cooler under the pump, let us put numbers on it, because 62 C on a 22 C floor means the unit is rejecting heat somewhere around 3.5 to 4 kilowatts steady state, and a tank is a storage tank, not a radiator; it sets the time constant of the heating, not the equilibrium temperature. I have asked our cooler supplier for the sizing worksheet on a 60 litre per minute flow, target 46 C return from the 62 C operating point with a 38 kW prime mover and a 12 kW installed pump power. The equations I want used are the standard ones, heat to be rejected equal to the product of installed power and the load factor minus what the reservoir wall sheds at 3 W per square metre kelvin on a plain steel tank, and I want the supplier to show me the numbers, not a catalogue page.
From: P. Lindqvist, Tank & Cooling Supplier — To: R. Almeida, D. Okafor; cc project manager
Subject: RE: HPU-07 — the sizing sheet, numbers first
Ricardo, the sheet is below, and the short answer is that your 62 C steady state is an equilibrium the tank itself has already accepted; you are not going to cool it by pouring in more steel. The 220-litre tank at 22 C ambient sheds roughly 1.1 kW through its surface by natural convection. Your circuit, at 12 kW installed pump power and a measured load factor of 0.45 on a two-shift press cycle, generates heat duty somewhere north of 4 kW at steady state. The difference, about 3 kW, is exactly what your 62 C equilibrium represents, because the tank simply settles wherever wall rejection balances generation. Doubling the tank to 440 litres adds surface and roughly 0.8 kW of extra shedding and stretches the warm-up time from forty minutes to ninety, but it does not change the equilibrium, the machine will still sit near 60 C, just slower to get there.
So the decision is a cooler, and here are parameters for your 55 C target at 22 C ambient: an air-oil cooler rated 6 kW at 60 C oil over 30 C ambient, running 60 litres per minute through a 90 kPa delta-P passage, sized plate-and-fin, no water to plumb, which matters because the customer’s floor has no cooling water at the press. With the cooler inline on the return, the tank becomes a settling vessel and a de-aerator again instead of a reluctant radiator, and 55 C is reachable with margin. Two design notes before you pick the model. First, one third of the tank volume must stay above the working level as air space, so on a 220-litre tank your usable oil to the pump is about 140 to 150 litres, which is tight against the old rule of two to three times pump flow, so I would extend the tank by 10 centimetres of height at the same footprint to give the cooler draw proper residence time. Second, put the return diffuser on the opposite end from the suction and keep the baffle, because a 6 kW cooler that dumps hot oil straight at the pump suction buys you nothing but a temperature probe reading that says cool while the pump heats.
From: H. Nakamura, Production — To: R. Almeida, P. Lindqvist; cc project manager, D. Okafor
Subject: RE: no budget line for a cooler, and the schedule does not move
Ricardo, I have to park the cooler on cost and schedule. The packaged price against the customer did not include a cooler, the pump room has no space beside the tank for a frame-mounted unit without moving the electrical cabinet, and the customer starts two-shift production in eleven days. I will own the decision criticism, but my read is we ship as is, at 62 C, and revisit at the next window, and I want the record to show the design team flagging it and the project accepting the risk under time pressure, because if the seals go in month three the conversation is going to be ugly and I want it on paper that this was a schedule call, not a design call.
From: S. Ferreira, Quality — To: H. Nakamura, R. Almeida; cc project manager
Subject: RE: oil temperature is a quality parameter, not a soft target
Hitoshi, with respect, the acceptance clause in the contract says 55 C at 40 C ambient, and we are at 62 C at 22 C. The viscosity drop matters, at 62 C a VG 46 oil sits around 34 to 38 cSt instead of the 46 cSt at 40 C the pumps were sized around, so leakage rises, and every seal and every accumulator bladder in that unit has a temperature budget that 62 C eats into on a two-shift duty. I am not asking to delay shipment over the geometry, I am asking that the 3 kW calculation, which nobody in this thread has challenged, gets resolved with a cooler now, because the cost of retrofitting a cooler on the customer site in eleven days is a crane and a shutdown, and the cost of fitting it in the pump room before we close the panels is one afternoon. The coolant madness is not the tank, and shipping a machine you know runs 7 degrees over spec to hit a date is a quality decision wearing a schedule costume.
From: D. Okafor, Customer Commissioning — To: R. Almeida; cc project manager
Subject: RE: brief — customer will accept a 4-week site cooler if documented
Ricardo, I spoke to the customer maintenance chief directly, and he is more reasonable than the contract suggests. He accepts a temporary operation at 62 C for up to four weeks provided three things are documented: an agreed cooler retrofit date, a viscosity check each week to prove the oil is not degrading, and a written heat-rejection calculation for the record. That takes the eleven-day gun away from the schedule and puts the decision back on engineering, where it belongs.
From: R. Almeida, Design — To: H. Nakamura, S. Ferreira, D. Okafor; cc P. Lindqvist, project manager
Subject: RE: decision — cooler ships with the machine, tank gains 100 mm
Decision, and it follows the numbers, not the org chart. We ship on schedule, and the 6 kW plate-and-fin air-oil cooler ships with the unit, installed in the pump room before the panels close, because the customer’s four-week window plus the cost comparison makes that strictly better than a site retrofit. The tank grows 100 millimetres in height at the same footprint, which lifts usable oil to about 175 litres and buys residence time for the cooler draw, and the diffuser stays opposite the suction behind the baffle. The cooler taps the return line after the filter, delta-P budget 90 kPa, and the fan is belt-driven off the same motor so there is no new start-up sequence for the operators to learn. We run the two-week soft-commissioning at 60 C maximum while the site cooler order sits in purchasing, but the acceptance run is the 55 C proof, and we have booked the test bay for it the Monday before shipment.
From: D. Okafor, Customer Commissioning — To: R. Almeida, H. Nakamura; cc project manager
Subject: RE: acceptance run passed at 52 C, log attached
Ran the acceptance four-hour cycle this morning, full press duty, 40 C equivalent ambient pulled to 32 C in the bay, and the probe sat at 52 C with the cooler fan on, 53 C with it cycling, comfortably inside the 55 C line with the margin you wanted. The warm-up now takes longer, about ninety minutes to equilibrium instead of forty, which the customer running two shifts will hardly notice, and the fishbone of the whole episode is simple: the tank set the time constant, the cooler set the equilibrium, and shipping a machine to 62 C to save a week would have cost us the acceptance.
From: R. Almeida, Design — To: all — Subject: RE: close-out, one table for the proposal folder
Closing HPU-07 as accepted. The numbers that survived the argument, they were not new, they were just not asked:
| Question | Rule of thumb | This unit |
|---|---|---|
| Tank volume | 2-3 x pump flow for air release | 220 L vs 60 L/min → borderline, raised 100 mm |
| Tank as cooler | sheds ~3 W/m2K nautral | 1.1 kW total, far below 4 kW duty |
| Equilibrium | set by heat balance, not tank size | 62 C without cooler, 52 C with |
| Viscosity guard | VG 46 target 38-46 cSt | 34-38 cSt at 62 C, margin restored |
Reference data — 60 L/min, 12 kW installed pump, load factor 0.45, 220 L tank → 175 L usable; wall shedding 1.1 kW; heat duty ~4 kW; 6 kW plate-and-fin air-oil cooler, 90 kPa delta-P; acceptance 55 C at 40 C ambient, measured 52-53 C at 32 C.
Glossary — heat duty: the steady-state heat the circuit must reject, roughly installed power times load factor; equilibrium temperature: where generation balances tank wall rejection, independent of tank volume; residence time: how long oil spends in the tank for air and dirt to settle; delta-P budget: the allowable pressure drop across a cooler placed in the return line; load factor: the fraction of installed power actually converted to heat in a duty cycle.
Field note — the tank that got sized by hope
A reservoir does not cool, it stores, and the moment a unit finds a warm equilibrium that is the heat balance speaking, not the tank being too small. Doubling the tank only stretches the warm-up and the invoice; the fix for a 62 C steady state is a cooler sized to the rejected kilowatts, which you get by multiplying installed power by the duty load factor and subtracting what the wall already sheds at about 3 W per square metre kelvin. Then honour the real tank job: enough usable volume, two to three times pump flow for aeration, a baffle, and the return diffuser on the far side from the suction. Compute the heat first, buy the cooler second, and keep the tank drawing honest, and the acceptance test stops being a gamble.
From: P. Lindqvist, Supplier — To: R. Almeida
Subject: RE: one handwritten appendix
Appendix from the bench to keep with the drawings: an oil cooler that sees particles before the return filter gets its fins scored and loses the 6 kW rating you bought, so keep the cooler downstream of the filter, and on a unit this size put the temperature probe in the tank, not in the cooler outlet, otherwise the controller reads a cool stream while the reservoir still bakes. Both were honoured on HPU-07, noted here for the next machine that inherits this conversation.
